You are on page 1of 8

326

Comparison of the automated Phoenix with the


Vitek 2 for the identification of Streptococcus
pneumoniae1
Scott A. Mittman, Richard C. Huard, Phyllis Della-Latta, and Susan Whittier

Abstract: Rapid and accurate identification of Streptococcus pneumoniae is a critical component in the optimal manage-
ment of infected patients. The performance of the BD Phoenix Automated Microbiology System (BD Diagnostic Systems,
Sparks, Md.) was evaluated for identification of S. pneumoniae (n = 311) and was compared to the Vitek 2 (bioMérieux,
Marcy l’Étoile, France). Strains with discordant identification between methods were resolved with 16S rRNA gene
sequencing as the gold standard. The Phoenix and the Vitek 2 correctly identified 96.8% (n = 301) and 95.2% (n = 296)
of S. pneumoniae strains, respectively. Overall, there was no statistically significant difference in the performance of the 2
automated systems for the identification of S. pneumoniae in this study. The Vitek 2 mean time-to-results for all strepto-
coccal identification was 1.5 h faster than that for the Phoenix. We conclude that the automated Phoenix and the Vitek 2
systems are comparable in their ability to identify S. pneumoniae and are preferable to the use of routine biochemical
assays, which have delayed time-to-results and are not dependably accurate.
Key words: identification, Streptococcus pneumoniae.
Résumé : L’identification rapide et fiable de Streptococcus pneumoniae est critique à la prise en charge optimale des pa-
tients infectés. La performance du système automatisé BD Phoenix (BD Diagnostic Systems, Sparks, Md.) a été évaluée
dans l’identification de S. pneumoniae (n = 311), et elle a été comparée à celle du système Vitek 2 (bioMérieux, Marcy
l’Étoile, France). Les souches dont les indentifications par les deux méthodes étaient discordantes ont été résolues par sé-
quençage de ARNr 16S comme utilisé comme étalon-or. Phoenix et Vitek 2 ont correctement identifié 96,8 % (n = 301) et
95,2 % (n = 296) des souches de S. pneumoniae, respectivement. Globalement, il n’y avait pas de différence significative
dans la performance des 2 systèmes automatisés pour identifier S. pneumoniae dans cette étude. Le temps moyen d’ob-
tention des résultats d’identification de tous les streptocoques du Vitek 2 était 1,5 h plus rapide que celui du Phoenix.
Nous concluons que les systèmes automatisés Phoenix et Vitek 2 sont comparables dans leur capacité à identifier S.
pneumoniae et qu’ils sont préférables à l’utilisation des essais biochimiques de routine, qui sont plus longs sans être plus
fiables pour autant.
Mots-clés : identification, Streptococcus pneumoniae.
[Traduit par la Rédaction]

Introduction Disease Control and Prevention (2008) recently reported


that 14% of hospitalized adults with invasive S. pneumoniae
Streptococcus pneumoniae infections are a significant do not survive their infections. Consequently, the accurate
cause of morbidity and mortality in the United States and in- and rapid identification of S. pneumoniae can impact
clude mucosal infections such as otitis media, conjunctivitis, optimal therapy and could improve patient outcomes.
and sinusitis, as well as invasive infections such as menin- Species-level identification of the a-hemolytic streptococci
gitis, bacteremia, and pneumonia. In the United States, S. through phenotypic differentiation can be particularly challeng-
pneumoniae is the leading cause of community-acquired ing, given both limited differential species-defining biochemi-
pneumonia and is the sixth most common cause of death cal characteristics and intra-species strain-to-strain biodiversity
due to an infectious disease (File 2004). The Centers for (Haanperä et al. 2007). In addition, some pneumococci may

Received 28 October 2009. Revision received 9 February 2010. Accepted 23 February 2010. Published on the NRC Research Press Web
site at cjm.nrc.ca on 16 April 2010.
S.A. Mittman. Clinical Microbiology Service, NewYork-Presbyterian Hospital, Columbia University Medical Center, 622 West 168th
Street, New York, NY 10032, USA.
R.C. Huard, P. Della-Latta, and S. Whittier.2 Clinical Microbiology Service, NewYork-Presbyterian Hospital, Columbia University
Medical Center, 622 West 168th Street, New York, NY 10032, USA; Department of Pathology, NewYork-Presbyterian Hospital,
Columbia University Medical Center, 622 West 168th Street, New York, NY 10032, USA.
1Portionsof this study were presented at the 2007 American Society of Microbiology 107th General Meeting in Toronto, Ontario,
Canada.
2Corresponding author (e-mail: whittie@nyp.org).

Can. J. Microbiol. 56: 326–332 (2010) doi:10.1139/W10-016 Published by NRC Research Press
Mittman et al. 327

yield misidentifications when single biochemical tests for a 0.5–0.6 McFarland standard of turbidity with a CrystalSpec
identification are used (Chandler et al. 2000; Whatmore et Nephelometer (BD Diagnostic Systems). Panels were sealed,
al. 2000; Kellogg et al. 2001; Pikis et al. 2001; Wester et logged, loaded into the Phoenix carousel, and incubated at
al. 2002; Verhelst et al. 2003; Arbique et al. 2004; Richter 35 8C. Data on colorimetric and fluorometric signals were
et al. 2008). Unfortunately, there is currently no phenotypic automatically collected by the instrument every 20 min and
‘‘gold standard’’ for accurate S. pneumoniae identification used to derive the kinetic measurements that provided the
(Kellogg et al. 2001; Haanperä et al. 2007), and genotypic ID determination. The Phoenix system ID database includes
identification methods are not widely utilized in most clini- S. pneumoniae and 31 other streptococcal species.
cal microbiology laboratories. For these reasons, improved
methods for the identification of S. pneumoniae are needed. Vitek 2
Automated systems are commonly used in clinical micro- Inoculation, reading, and interpretation of Vitek 2 panels
biology laboratories for routine pathogen identification were performed according to the manufacturer’s instructions.
because they generally offer benefits such as enhanced The Vitek 2 is an automated system for identification and
reproducibility, accuracy, and expedited time-to-results antimicrobial susceptibility testing using sealed disposable
(TTR). However, they are not often used for S. pneumoniae cards. The instrument detects bacterial growth and metabolic
identification because of system limitations in the identifica- reactions in the microwells of plastic test cards based on
tion of a-hemolytic streptococci and laboratory over-reliance colorimetric changes. Isolates were subcultured onto BAP
on traditional biochemical tests. In this study, we performed and incubated for 24 h at 37 8C before testing on the Vitek
a side-by-side comparison of 2 automated systems, the BD 2 system. The inoculum was prepared by adjusting a
Phoenix Automated Microbiology System (BD Diagnostic bacterial suspension to a McFarland standard of 0.5–0.63 in
Systems, Sparks, Md.) and the Vitek 2 (bioMérieux, Marcy a 0.45% sodium chloride solution. The Vitek 2 ID database
l’Étoile, France), to evaluate their ability to correctly includes 33 streptococcal species, including S. pneumoniae
identify S. pneumoniae. (Haanperä et al. 2007).

Materials and methods Interpretation of results


Study isolates Isolate ID results were divided into one of 4 categories: cor-
This comparative evaluation studied 311 unique S. pneu- rect ID; low level of discrimination (inability to differentiate
moniae clinical isolates. Each single isolate was recovered between 2 species); misidentification (incorrect assignment);
from individual patient specimens at a major academic med- and no ID (no assignment).
ical center in New York City over a 4-year period (2003–
2007). Respiratory sources accounted for 68% of the total Resolution of ID discrepancies
isolates (212 of 311), 36 isolates were derived from blood IDs other than S. pneumoniae generated by either the
(11.6%), 20 isolates (6.4%) from the eye, and 43 isolates Phoenix system or the Vitek 2 resulted in repeat testing.
from other sources (e.g., wound, ear, tissue, body fluid). Identification was repeated in parallel on both automated
The test pool of isolates was a combination of prospectively systems alongside routine biochemical testing to confirm
acquired (n = 167) and frozen strains (n = 144). All isolates results. Isolates that were not identified as S. pneumoniae
used in the study had been previously identified as S. pneu- by one or more methods were subjected to 16S rRNA
moniae by biochemical tests. sequencing for species determination.

Routine culture and biochemical-based identification Amplification and sequencing of the 16S rRNA gene
Streptococcus pneumoniae isolates were identified ini- Bacterial DNA was extracted from pure cultures using the
tially by colony morphology, hemolysis on BBL Columbia BioRobot EZ1 Magtration System 68G (Qiagen, Basel,
Agar with 5% sheep blood agar (BAP) (BD Diagnostic Sys- Switzerland) and the Qiagen EZ1 DNA Blood 350 mL kit.
tems), slide co-agglutination (Phaedebact, Remel, Lenexa, Amplification, sequencing, and in silico analysis of the resultant
Kans.), optochin susceptibility (P-disk containing 5 mg of ~1533 bp streptococcal 16S rRNA PCR fragment were
optochin (BD Diagnostic Systems) and (or) bile solubility performed as previously described (Schlaberg et al. 2008).
(BD Diagnostic Systems) testing, following the instructions Species-level identification was accomplished by both
of the manufacturer. The species identity of a random subset BLASTN analysis of full-length consensus sequences in
of typical S. pneumoniae strains (n = 10) were confirmed by GenBank (http://www.ncbi.nlm.nih.gov/BLAST/) as well as
16S rRNA gene sequencing. by comparing against 16S rRNA sequences in an in-house
database. The 16S rRNA database was derived from the
Phoenix sequencing of control and QC streptococcal strains of
The identification of S. pneumoniae using the Phoenix known identity and downloads from the complete genomic
system was performed according to the manufacturer’s pro- sequence of S. pneumoniae strain TIGR4 (GenBank
tocols (Donay et al. 2004; Kanemitsu et al. 2005). Briefly, accession No. AE005672). Characterization of the nearly
Phoenix uses an identification (ID) and antimicrobial sus- full-length 16S rRNA gene sequence (Petti 2007) by this
ceptibility testing combination panel (SMIC/ID-100), with protocol allows for the differentiation of most streptococcal
the ID substrates located on one side and the microwells for species, including those with infrequent species-specific poly-
AST on the other. Following 18–24 h of incubation on BAP, morphisms, such as those that occur between S. pneumoniae
a bacterial suspension was made in ID broth and adjusted to and Streptococcus mitis (our unpublished data).

Published by NRC Research Press


328 Can. J. Microbiol. Vol. 56, 2010

Performance and statistical analyses Table 1. Comparison of automated systems for the iden-
The statistical significance of differences in the ID results tification (ID) of Streptococcus pneumoniae (n = 311).
between the automated systems was calculated using Fish-
No. of strains (%)
er’s exact test. Results were deemed significant with a p £
0.05. Phoenix Vitek 2
Correct ID
Results Species-level ID 293 (94.2) 296 (95.2)
Low-level discrimination 8 (2.6) 0.
Identification of S. pneumoniae isolates Total 301 (96.8) 296 (95.2)
Comparative results obtained for the identification of 311
Discordant ID
S. pneumoniae isolates by the Phoenix and the Vitek 2 are
Misidentification 4 (1.3) 13 (4.2)
summarized in Table 1. The Phoenix produced an accurate
No ID 6 (1.9) 2 (0.6)
identification for 293 S. pneumoniae strains (94.2%), while
Total 10 (3.2) 15 (4.8)
another 8 S. pneumoniae strains (2.6%) had low-level
discrimination IDs (S. mitis/S. pneumoniae) that were
resolved by supplementary bile solubility testing, as sug- Vitek 2 provided discordant IDs for 4 S. pneumoniae
gested by the Phoenix expert system. By comparison, the strains in common. Overall, for difficult-to-identify S.
Vitek 2 correctly identified 296 strains (95.2%); there were pneumoniae isolates, the Phoenix tended to provide no ID
no strains with low-level discrimination IDs. Overall, 301 (6 of 10 strains) while the Vitek 2 was more likely to re-
(96.8%) and 296 (95.2%) S. pneumoniae isolates were port an incorrect ID (13 of 15 strains) (p = 0.028, Fisher’s
correctly identified by the Phoenix and Vitek 2, respectively.
exact test).
With regard to biochemical testing, 304 of the 311 isolates
Interestingly, 7 S. pneumoniae strains that were difficult
(98.8%) were typical pneumococci, and 7 were considered
to identify using one or both of the automated systems also
atypical with variable biochemical test results. The propor-
showed unexpected variability in biochemical testing
tion of isolates with biochemical test results consistent with
(Table 2). One strain was slide co-agglutination negative
those expected for S. pneumoniae were 98.4% (306/311),
and optochin resistant, 2 strains were bile insoluble, 2 strains
98.7% (307/311) and 99.7% (310/311) for optochin, bile sol-
were bile insoluble and optochin resistant, and 2 strains
ubility, and slide co-agglutination, respectively. In terms of
were optochin resistant. Such variability in biochemical test-
discordant IDs, the Phoenix resulted in 4 misidentifications
and 6 no IDs (3.2% in total) compared to the Vitek 2, with ing may be suggestive of broader phenotypic heterogeneity
13 misidentifications and 2 no IDs (4.8% in total). Of the 7 and thereby correlate with the misidentification of these
biochemically ‘‘atypical’’ pneumococci, 3 were correctly strains by the automated systems.
identified by Phoenix and 2 by Vitek 2. Two ‘‘atypical’’ Overall, the strains with low-level discrimination and dis-
isolates were incorrectly identified by both systems. cordant results were recovered largely from sputum (n = 9)
Although there were fewer misidentifications of pneumo- and eye sources (n = 9), accounting for 78% (18/23) of the
cocci by the Phoenix than by the Vitek 2, there was no 23 isolates described in Table 2. Tracheal aspirates (n = 2),
statistically significant difference in the performance of the throat (n = 2), and blood (n = 1) accounted for the remain-
2 automated systems for the identification of S. pneumoniae der.
strains in this study (p = 0.415, Fisher’s exact test).
Analysis of Vitek 2 substrate patterns for S. pneumoniae
Analysis of S. pneumoniae ID discrepancies discordants
The true IDs for discordant isolates were investigated Ten of 13 (77.0%) S. pneumoniae misidentifications by
using 16S rRNA sequencing, and routine biochemical results the Vitek 2 were designated as non-streptococci. Table 3
were confirmed with repeat testing. Confirmatory testing lists select individual Vitek 2 biochemical tests and the
showed that all atypical biochemical results were the same results that were obtained for the misidentified strains.
when repeated. Table 2 lists the evaluation results of each These strains were variably negative for 9 tests that were
16S rRNA sequence-confirmed S. pneumoniae strain that expected to be positive for S. pneumoniae. In addition, 9
produced a low-level discrimination or discordant ID by the strains (reported as Granulicatella adiacens) were positive
Phoenix and (or) the Vitek 2 as compared to the results of for the a-glucosidase test, a result that was inconsistent
repeat biochemical testing. The data showed that Vitek 2 with a Granulicatella adiacens ID, but was an expected re-
either misidentified (n = 10) (as Granulicatella adiacens, sult for S. pneumoniae. Similarly, another strain (resulted
Gardnerella vaginalis, S. mitis/Streptococcus oralis, or as Gardnerella vaginalis) was positive for 2 tests (tyrosine
Streptococcus pasteurianus) or reported no ID (n = 1) for arylamidase and D-maltose) that were not indicative of
S. pneumoniae strains for which Phoenix either provided a Gardnerella vaginalis but rather of S. pneumoniae. There
correct species-level ID (n = 5) or reported a low-level dis- were also 3 misidentified strains that were correctly identi-
crimination ID (n = 6). In addition, there were 2 low-level fied to the genus level, but with an incorrect species assign-
discrimination IDs by the Phoenix that were correctly iden- ment, as S. mitis/S. oralis (n = 2) and S. pasteurianus (n =
tified by Vitek 2 as S. pneumoniae and 6 strains that the 1). Each of these was negative for L-pyrrolidonyl arylami-
Phoenix either misidentified as S. mitis (n = 3) or S. oralis dase, while the S. pasteurianus strain was additionally
(n = 1) or provided no ID (n = 2) that the Vitek 2 negative for b-galactopurinosidase, which was consistent
correctly identified as S. pneumoniae. Both Phoenix and with the subsequent test ID result. Therefore, unexpected

Published by NRC Research Press


Mittman et al. 329

Table 2. Analysis of low-level discrimination and discordant Streptococcus pneumoniae isolatesa (n = 23).

Automated systems Biochemical testsb


No. of
strains Phoenix Vitek 2 Co-agglutination Bile solubility Optochin
1 S. mitis/S. pneumoniae Granulicatella adiacens – + R
4 S. mitis/S. pneumoniae Granulicatella adiacens + + S
1 S. mitis/S. pneumoniae S. pasteurianus + + R
2 S. mitis/S. pneumoniae S. pneumoniae + + S
1 S. pneumoniae Granulicatella adiacens + + S
1 S. pneumoniae Gardnerella vaginalis + + S
1 S. pneumoniae No ID + – S
2 S. pneumoniae S. mitis/S. oralis + + S
1 S. mitis S. pneumoniae + – S
2 S. mitis S. pneumoniae + + S
1 S. oralis S. pneumoniae + – R
2 No ID S. pneumoniae + + S
1 No ID Granulicatella adiacens + + R
2 No ID Granulicatella adiacens + + S
1 No ID No ID + – R
a
Confirmed by 16S rRNA sequencing.
b
Expected results: slide co-agglutination (+); bile solubility (+); optochin (S). Bile solubility: +, positive; –, negative. Optochin
susceptibility: S, susceptible, R, resistant.

Table 3. Vitek 2 discordant identification biochemical analysisa (n = 13).

No. of
strains Vitek 2 ID AGLU TyrA BGAR BGAL dMAL SAC dGAL dTRE LAC NAG PyrA
5 Granulicatella adiacens + + – – – – – – – – –
3 Granulicatella adiacens + + – – – – – – – – +
1 Granulicatella adiacens + + – – + + – – – + +
1 Gardnerella vaginalis + + + + + – – – – – –
2 S. mitis/S. oralis + + + + + + + + + + –
1 S. pasteurianus + + – + + + + + + + –
Note: AGLU, a-glucosidase; TyrA, tyrosine arylamidase; BGAR, b-galactopurinosidase; BGAL, b-galactosidase; dMAL, D-maltose; SAC, saccharose–
sucrose; dGAL, D-galactosidase; dTRE, D-trehalose; LAC, lactose; NAG, N-acetyl D-glucosamine; PyrA, L-pyrrolidonyl arylamidase.
a
Expected results for Streptococcus pneumoniae were (+) for all biochemicals.

biopatterns of substrate utilization appeared to be the cause in their ability to correctly identify S. pneumoniae (96.8% vs.
of the incorrect identification by the Vitek 2. 95.2%, respectively). However, the Phoenix did report fewer
S. pneumoniae discrepant IDs (3.2%) than that reported by
Time to results for identification of S. pneumoniae the Vitek 2 (4.8%), but these differences were not statisti-
The TTR for the Phoenix system ranged from 2.16 to cally significant. The correct identification of 96.8% of S.
12.48 h. Overall, the Phoenix system averaged 6.28 h, with pneumoniae isolates by the Phoenix that we reported is in
5.94 h for concordant isolates, 8.3 h for misidentification general agreement with the reported statistics of 85.9% by
isolates, and 12.35 h for low-level discrimination IDs (S. Kanemitsu et al. (2005), 93.8% by Hirakata et al. (2005),
mitis/S. pneumoniae). Vitek 2 TTR ranged from 2.25 to and 100% by Brigante et al. (2006). However, it should be
8.5 h and averaged 4.50 h for all isolates. The TTR average noted that each of these studies examined isolate pools of
for a concordant ID was 4.46 and 5.75 h for misidentifica- S. pneumoniae of a considerably smaller size (n = 92, 97,
tions. The Vitek 2 mean time for all IDs was therefore 28% and 46, respectively) than the 311 S. pneumoniae isolates
(~1.5 h) faster than the mean Phoenix ID time. that were evaluated herein.
In this study, there were several S. pneumoniae strains
that proved difficult for one or both of the automated
Discussion systems to correctly or definitively identify. For instance,
To the best of our knowledge, there are no reports com- there were 8 S. pneumoniae strains for which Phoenix
paring the test characteristics of the Phoenix system against reported a low-level discrimination ID. It should be noted
the Vitek 2 that specifically focus on identification of S. that in each of the low-level discrimination IDs we
pneumoniae. In this study, these 2 methods were comparable described, the Phoenix suggested supplementary tests that

Published by NRC Research Press


330 Can. J. Microbiol. Vol. 56, 2010

confirmed a S. pneumoniae ID, rather than generating a de- sensitivities and high specificities (Kellogg et al. 2001;
finitive incorrect ID. Six of the aforementioned Phoenix Richter et al. 2008; Werno and Murdoch 2008). However,
nondiscriminatory ID isolates were a source of difficulty there are increasing reports of atypical S. pneumoniae and
with the Vitek 2 as well, but in contrast to the Phoenix, the S. mitis with unusual patterns of bile solubility and (or)
Vitek 2 produced an incorrect ID (Granulicatella adiacens optochin susceptibility that might make their identification
(n = 5) or S. pasteurianus (n = 1)). In terms of the Phoenix and differentiation challenging (Phillips et al. 1988; Fenoll
S. pneumoniae misidentifications that were observed, there et al. 1990; Mundy et al. 1998; Kearns et al. 2000; What-
were 4 strains that were reported as S. mitis (n = 3) and S. more et al. 2000; Kellogg et al. 2001; Pikis et al. 2001;
oralis (n = 1), but that were correctly identified by the Vitek Kaijalainen et al. 2002; Wester et al. 2002; Ko et al. 2005;
2. It is noteworthy that S. pneumoniae is a member of the S. Richter et al. 2008). In the course of routine clinical labora-
mitis group, which includes S. mitis, S. oralis, Streptococcus tory testing coincident with this study, we encountered 9
pseudopneumoniae, and Streptococcus sanguinis. The chal- streptococcal strains that could not be definitively identified
lenge posed by the aforementioned species to the automated or excluded as S. pneumoniae on the basis of biochemical
systems might therefore be explained by their close genetic testing alone. That is, these strains gave mixed patterns of re-
relationship, inferred from the high interspecies 16S rRNA sults for slide co-agglutination, bile solubility or insolubility,
sequence similarity that exists among members of the S. mi- and optochin susceptibility or resistance (data not shown).
tis group (Facklam 2002; Woo et al. 2003; Arbique et al. Sequencing of 16S rRNA determined that 8 strains were S.
2004). The 16S rRNA of S. pneumoniae and S. mitis, for mitis and 1 strain was S. pseudopneumoniae. Similar obser-
example, are >99% identical (Spellerberg and Brandt 2007). vations have been made by others where the reliance on
As the Phoenix and Vitek 2 systems utilize ID paradigms ‘‘classic criteria’’ for S. pneumoniae ID has been suggested
that rely upon expected biopatterns, the potential exists for to occasionally result in the misidentification of orophar-
biovariable strains that will confound identification. In this yngeal streptococci as S. pneumoniae (Wester et al. 2002).
study, there were 6 S. pneumoniae isolates that were not Indeed, Richter et al. (2008) showed that approximately
identified by the Phoenix, 4 of 6 of which were also prob- 5% of S. pneumoniae isolates, collected from 41 US
lematic with Vitek 2, with 1 strain not being identified and medical centers for surveillance purposes, were not pneu-
3 others assigned to Granulicatella adiacens. With each of mococci. Moreover, given the high percentage of penicillin
the Vitek 2 S. pneumoniae misidentifications, the ID confi- resistance with S. mitis (Spellerberg and Brandt 2007) and
dence percentages were high (data not shown), thus giving S. oralis (Facklam 2002), the subsequent overreporting of
the false impression of an accurate final result. Overall, the penicillin-resistant S. pneumoniae could result (Haanperä et
Vitek 2 incorrectly identified 9 S. pneumoniae strains as al. 2007). This has clinical relevance owing to the potential
Granulicatella adiacens (n = 9), a bacterium belonging to a consequence of unnecessary changes in antibiotic prescrip-
genus also known as nutritionally deficient streptococci tion practices, as truly penicillin-resistant S. pneumoniae
(Robson et al. 2007) and 1 other as Gardnerella vaginalis. isolates are also recognized to be less susceptible to mac-
Looking more closely at these identifications and their rolides, cephalosporins, and trimethoprim–sulfamethoxazole
respective substrate utilization patterns (Table 3), these (Patel et al. 2000).
isolates could have been identified by default because of At present, optochin susceptibility appears to be the phe-
poor growth within the microwells on the Vitek 2 GP card. notypic characteristic most frequently used to differentiate
In fact, for most of the misidentified strains, there were sub- between S. pneumoniae and other a-hemolytic streptococci
strate utilization signals that were counterindicative of the (Chandler et al. 2000; Pikis et al. 2001; Verhelst et al.
Vitek 2 assignment and yet agreed with the S. pneumoniae 2003; Bosshard et al. 2004) and oftentimes is used singu-
expected biopattern. Abele-Horn et al. (2006) reported larly for identification. However, estimates of optochin
similar misidentifications with Vitek 2, where they found resistance among S. pneumoniae are given between 1%
Granulicatella adiacens and Gardnerella vaginalis named 6 (Robson et al. 2007) and 4% (Spellerberg and Brandt
times out of 162 pneumococci tested. According to another 2007), and examples have been noted in reports going back
study from Woo et al. (2003), Vitek 2 was unable to identify at least 20 years (Kontiainen and Sivonen 1987; Phillips et
a single Granulicatella adiacens isolate (of 7). Therefore, al. 1988). In fact, were our differentiation based upon optochin
resulted by the Vitek 2, Granulicatella adiacens should be testing alone, 5 of the 16S rRNA sequencing-confirmed S.
viewed with a high degree of suspicion and be confirmed pneumoniae strains (1.6%) studied here (Table 2) would have
by supplemental testing. In an attempt to improve identifica- been misidentified.
tion success with the Vitek 2, 12 of the problematic strains Optochin-sensitive oral nonpneumococcal streptococci are
were retested at an adjusted bacterial suspension of 0.65–0.7 increasingly reported as well (Whatmore et al. 2000; Pikis et
McFarland turbidity (from 0.5–0.63). By increasing the al. 2001). For example, Balsalobre et al. (2006) described
inoculum density, we were able to improve the ID concord- optochin susceptibility for S. mitis, a microorganism that is
ance with the Vitek 2 from 95.2% to 96.8% (data not commonly understood to be resistant. Although statistics
shown), resolving 5 of the discrepant Granulicatella regarding bile-insoluble S. pneumoniae are lacking, 4 of the
adiacens IDs. This modification in protocol might therefore 16S rRNA sequencing-confirmed S. pneumoniae strains
be helpful in resolving similar unexpected Vitek 2 calls in (1.3%) we evaluated gave an unexpected bile insolubility
the future. test result (Table 2). On the other hand, other biochemical
Bile solubility, optochin susceptibility, and serological tests used to detect the presence of capsular antigens, such
methods of S. pneumoniae antigen detection are widely as by slide co-agglutination, have been reported to have
used assays that are generally reported to have very high very high specificities and sensitivities (Kellogg et al. 2001)

Published by NRC Research Press


Mittman et al. 331

and might be more appropriate for ID confirmation. In fact, pneumoniae disease [online]. Available from http://www.cdc.
based on our observations (Table 2), slide co-agglutination gov/ncidod/dbmd/diseaseinfo/streppneum_t.htm [accessed 9 Sep-
was the best biochemical test for streptococcal differentiation, tember 2009].
where there was only 1 instance of negative co-agglutination Chandler, L.J., Reisner, B.S., Woods, G.L., and Jafri, A.K. 2000.
for a 16S rRNA sequencing-confirmed S. pneumoniae isolate Comparison of four methods for identifying Streptococcus pneu-
(0.3%). Therefore, the overall data appear to suggest that it moniae. Diagn. Microbiol. Infect. Dis. 37(4): 285–287. doi:10.
is insufficient to rely upon a single biochemical test method 1016/S0732-8893(00)00149-8. PMID:10974582.
to differentiate atypical S. pneumoniae strains from viridans Donay, J.L., Mathieu, D., Fernandes, P., Prégermain, C., Bruel, P.,
streptococci. Wargnier, A., et al. 2004. Evaluation of the automated phoenix
system for potential routine use in the clinical microbiology la-
For clinical laboratories considering the implementation of
boratory. J. Clin. Microbiol. 42(4): 1542–1546. doi:10.1128/
either of these automated systems, they would be most cost
JCM.42.4.1542-1546.2004. PMID:15071001.
effective for laboratories with sizeable throughput. While Facklam, R. 2002. What happened to the streptococci: overview of
smaller laboratories could be discouraged by a number of taxonomic and nomenclature changes. Clin. Microbiol. Rev.
factors (i.e., cost, dedicated footprint, strict ambient temper- 15(4): 613–630. doi:10.1128/CMR.15.4.613-630.2002. PMID:
ature operating range, technologist training, and computer 12364372.
literacy requirement), attributes such as an advanced expert Fenoll, A., Martinez-Suarez, J.V., Muñoz, R., Casal, J., and Garcia,
system, the ability to process large numbers of isolates, rapid J.L. 1990. Identification of atypical strains of Streptococcus
turnaround time, accuracy, reproducibility, and the ability to pneumoniae by a specific DNA probe. Eur. J. Clin. Microbiol.
track results, along with the ease of availability of data for Infect. Dis. 9(6): 396–401. doi:10.1007/BF01979468. PMID:
epidemiological monitoring, are likely to have a favorable 2387292.
impact on the laboratory workflow of mid- to large-sized File, T.M. 2004. Streptococcus pneumoniae and community-acquired
laboratories with concomitant high workload volume. pneumonia: a cause for concern. Am. J. Med. Suppl. 117(3, Suppl.
In conclusion, the present study demonstrated comparable 1): 39–50. doi:10.1016/j.amjmed.2004.07.007.
performance between the Phoenix and Vitek 2 for identifica- Haanperä, M., Jalava, J., Huovinen, P., Meurman, O., and
tion of S. pneumoniae. In addition, our findings underscored Rantakokko-Jalava, K. 2007. Identification of a-hemolytic strep-
the accuracy and advantage of using automated systems for tococci by pyrosequencing the 16S rRNA gene and by use of VI-
S. pneumoniae identification and reserving biochemical TEK 2. J. Clin. Microbiol. 45(3): 762–770. doi:10.1128/JCM.
testing for supplemental evaluation. 01342-06. PMID:17215341.
Hirakata, Y., Matsuda, J., Nakano, M., Hayashi, T., Tozaka, S.,
Acknowledgements Takezawa, T., et al. 2005. Evaluation of the BD Phoenix Auto-
mated Microbiology System SMIC/ID panel for identification
This study was supported in part by BD Diagnostic Systems. and antimicrobial susceptibility testing of Streptococcus spp. Di-
Vitek 2 cards were provided by bioMérieux. agn. Microbiol. Infect. Dis. 53(3): 169–173. doi:10.1016/j.
diagmicrobio.2005.06.003. PMID:16243478.
References Kaijalainen, T., Rintamäki, S., Herva, E., and Leinonen, M. 2002.
Abele-Horn, M., Stoy, K., Frosch, M., and Reinert, R.R. 2006. Evaluation of gene-technological and conventional methods in
Comparative evaluation of a new Vitek 2 system for identifica- the identification of Streptococcus pneumoniae. J. Microbiol.
tion and antimicrobial susceptibility testing of Streptococcus Methods, 51(1): 111–118. doi:10.1016/S0167-7012(02)00061-1.
pneumoniae. Eur. J. Clin. Microbiol. Infect. Dis. 25(1): 55–57. PMID:12069896.
doi:10.1007/s10096-005-0071-1. PMID:16424974. Kanemitsu, K., Kunishima, H., Inden, K., Hatta, M., Harigae, H.,
Arbique, J.C., Poyart, C., Trieu-Cuot, P., Quesne, G., Carvalho, Ishizawa, K., and Kaku, M. 2005. Evaluation of the BD Phoenix
M.G., Steigerwalt, A.G., et al. 2004. Accuracy of phenotypic SMIC/ID, a new streptococci identification and antimicrobial
and genotypic testing for identification of Streptococcus pneu- susceptibility panel, for potential routine use in a university-
moniae and description of Streptococcus pseudopneumoniae sp. based clinical microbiology laboratory. Diagn. Microbiol. Infect.
nov. J. Clin. Microbiol. 42(10): 4686–4696. doi:10.1128/JCM. Dis. 53(2): 101–105. doi:10.1016/j.diagmicrobio.2005.05.016.
42.10.4686-4696.2004. PMID:15472328. PMID:16168613.
Balsalobre, L., Hernández-Madrid, A., Llull, D., Martı́n-Galiano, Kearns, A.M., Wheeler, J., Freeman, R., Seiders, P.R., Perry, J.,
A.J., Garcı́a, E., Fenoll, A., and de la Campa, A.G. 2006. Mole- Whatmore, A.M., and Dowson, C.G. 2000. Pneumolysin detec-
cular characterization of disease-associated streptococci of the tion identifies atypical isolates of Streptococcus pneumoniae. J.
mitis group that are optochin susceptible. J. Clin. Microbiol. Clin. Microbiol. 38(3): 1309–1310. PMID:10755884.
44(11): 4163–4171. doi:10.1128/JCM.01137-06. PMID: Kellogg, J.A., Bankert, D.A., Elder, C.J., Gibbs, J.L., and Smith,
16971639. M.C. 2001. Identification of Streptococcus pneumoniae revis-
Bosshard, P.P., Abels, S., Altwegg, M., Böttger, E.C., and Zbinden, ited. J. Clin. Microbiol. 39(9): 3373–3375. doi:10.1128/JCM.39.
R. 2004. Comparison of conventional and molecular methods 9.3373-3375.2001. PMID:11526182.
for identification of aerobic catalase-negative gram-positive Ko, K.S., Oh, W.S., Peck, K.R., Lee, J.H., Lee, N.Y., and Song,
cocci in the clinical laboratory. J. Clin. Microbiol. 42(5): 2065– J.H. 2005. Phenotypic and genotypic discrepancy of Streptococ-
2073. doi:10.1128/JCM.42.5.2065-2073.2004. PMID:15131171. cus pneumoniae strains isolated from Asian countries. FEMS
Brigante, G., Luzzaro, F., Bettaccini, A., Lombardi, G., Meacci, F., Immunol. Med. Microbiol. 45(1): 63–70. doi:10.1016/j.femsim.
Pini, B., et al. 2006. Use of the Phoenix automated system for 2005.01.012. PMID:15985224.
identification of Streptococcus and Enterococcus spp. J. Clin. Kontiainen, S., and Sivonen, A. 1987. Optochin resistance in Strep-
Microbiol. 44(9): 3263–3267. doi:10.1128/JCM.00299-06. tococcus pneumoniae strains isolated from blood and middle ear
PMID:16954258. fluid. Eur. J. Clin. Microbiol. Infect. Dis. 6(4): 422–424. doi:10.
Centers for Disease Control and Prevention. 2008. Streptococcus 1007/BF02013101.

Published by NRC Research Press


332 Can. J. Microbiol. Vol. 56, 2010

Mundy, L.S., Janoff, E.N., Schwebke, K.E., Shanholtzer, C.J., and Clin. Microbiol. 46(1): 265–273. doi:10.1128/JCM.00937-07.
Willard, K.E. 1998. Ambiguity in the identification of Strepto- PMID:18003809.
coccus pneumoniae: optochin, bile solubility, quellung, and the Spellerberg, B., and Brandt, C. 2007 Streptococcus. In Manual of
AccuProbe DNA probe tests. Am. J. Clin. Pathol. 109(1): 55– clinical microbiology. Vol. 1. Edited by P.R. Murray, E.J.
61. PMID:9426518. Baron, J.H. Jorgensen, M.L. Landry, and M.A. Pfaller. ASM
Patel, R., Rouse, M.S., Piper, K.E., and Steckelberg, J.M. 2000. In Press, Washington, D.C. pp. 412–429.
vitro activity of GAR-936 against vancomycin-resistant entero- Verhelst, R., Kaijalainen, T., De Baere, T., Verschraegen, G.,
cocci, methicillin-resistant Staphylococcus aureus and penicil- Claeys, G., Van Simaey, L., et al. 2003. Comparison of five
lin-resistant Streptococcus pneumoniae. Diagn. Microbiol. genotypic techniques for identification of optochin-resistant
Infect. Dis. 38(3): 177–179. doi:10.1016/S0732-8893(00)00193- pneumococcus-like isolates. J. Clin. Microbiol. 41(8): 3521–
0. PMID:11109018. 3525. doi:10.1128/JCM.41.8.3521-3525.2003. PMID:12904349.
Petti, C.A. 2007. Detection and identification of microorganisms by Werno, A.M., and Murdoch, D.R. 2008. Medical microbiology: la-
gene amplification and sequencing. Clin. Infect. Dis. 44(8): boratory diagnosis of invasive pneumococcal disease. Clin. In-
1108–1114. doi:10.1086/512818. PMID:17366460. fect. Dis. 46(6): 926–932. doi:10.1086/528798. PMID:18260752.
Phillips, G., Barker, R., and Brogan, O. 1988. Optochin-resistant Wester, C.W., Ariga, D., Nathan, C., Rice, T.W., Pulvirenti, J.,
Streptococcus pneumoniae. Lancet, 332(8605): 281–281. doi:10. Patel, R., et al. 2002. Possible overestimation of penicillin resis-
1016/S0140-6736(88)92573-1. PMID:2899267. tant Streptococcus pneumoniae colonization rates due to misi-
Pikis, A., Campos, J.M., Rodriguez, W.J., and Keith, J.M. 2001. dentification of oropharyngeal streptococci. Diagn. Microbiol.
Optochin resistance in Streptococcus pneumoniae: mechanism, Infect. Dis. 42(4): 263–268. doi:10.1016/S0732-8893(01)00358-
significance, and clinical implications. J. Infect. Dis. 184(5): 3. PMID:12007444.
582–590. doi:10.1086/322803. PMID:11474432. Whatmore, A.M., Efstratiou, A., Pickerill, A.P., Broughton, K.,
Richter, S.S., Heilmann, K.P., Dohrn, C.L., Riahi, F., Beekmann, Woodard, G., Sturgeon, D., et al. 2000. Genetic relationships be-
S.E., and Doern, G.V. 2008. Accuracy of phenotypic methods tween clinical isolates of Streptococcus pneumoniae, Streptococ-
for identification of Streptococcus pneumoniae isolates included cus oralis, and Streptococcus mitis: characterization of
in surveillance programs. J. Clin. Microbiol. 46(7): 2184–2188. ‘‘Atypical’’ pneumococci and organisms allied to S. mitis har-
doi:10.1128/JCM.00461-08. PMID:18495854. boring S. pneumoniae virulence factor-encoding genes. Infect.
Robson, R.L., Essengue, S., Reed, N.A., and Horvat, R.T. 2007. Immun. 68(3): 1374–1382. doi:10.1128/IAI.68.3.1374-1382.
Optochin resistance in Streptococcus pneumoniae induced by 2000. PMID:10678950.
frozen storage in glycerol. Diagn. Microbiol. Infect. Dis. 58(2): Woo, P.C.Y., Fung, A.M.Y., Lau, S.K.P., Chan, B.Y.L., Chiu, S.K.,
185–190. doi:10.1016/j.diagmicrobio.2007.01.006. PMID: Teng, J.L.L., et al. 2003. Granulicatella adiacens and Abiotro-
17350207. phia defectiva bacteraemia characterized by 16S rRNA gene se-
Schlaberg, R., Huard, R.C., and Della-Latta, P. 2008. Nocardia cy- quencing. J. Med. Microbiol. 52(2): 137–140. doi:10.1099/jmm.
riacigeorgica, an emerging pathogen in the United States. J. 0.04950-0. PMID:12543919.

Published by NRC Research Press


Copyright of Canadian Journal of Microbiology is the property of NRC Research Press and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.

You might also like