Difficulties in Identifying Klebsiella Strains of Clinical Origin

Zbl. Bakt. 274, 456-464 (1991) © Gustav Fischer Verlag, StuttgartlNew York

Difficulties in Identifying Klebsiella Strains of Clinical Origin DOMINIQUE MONNET l , JEAN FRENEYZ, YVONNE BRUNI, JEAN-MARC BOEUFGRAS 3 , and JEAN FLEURETTE l Laboratoire de Bacteriologie-Virologie, Faculte de Medecine Alexis Carrel 69372 Lyon Cedex 08, France 2 Laboratoire de Bacteriologie, Faculte de Pharmacie, 69373 Lyon Cedex 08, France 3 API-Unite Bacteriologie bioMerieux, La Balme-les-Grottes, 38390 Montalieu-Vercieu, France 1

With 1 Figure· Received April 21, 1990 . Accepted in revised form August 1, 1990

Summary Two hundred and four strains of Gram-negative bacteria of clinical origin, initially identified as Klebsiella using the API 20 E system, and 10 reference strains were further analysed with the API 20 EC test system and the API 50 CH, API 50 AO, API 50 AA assimilation systems. Four clusters corresponding to the species Klebsiella pneumoniae subsp. pneumoniae, K. oxytoca, K. planticola, and K. terrigena were formed after numerical analysis of 155 selected tests and the 26 most discriminating tests were determined. A comparison was made between conventional identification using the API 20 E system and the results of the numerical analysis. The conventional method resulted in incorrect identification of 13% of the strains tested, especially for the new species: K. planticola and K. terrigena. After numerical analysis, 17 out of 204 strains (8.3 %) of clinical origin were identified as K. planticola. Only 1 strain of clinical origin was identified as K. terrigena, and 1 strain as K. ornithinolytica. Zusammenfassung 204 Stamme gram-negativer Bakterien klinischen Ursprungs, die anfanglich unter Verwendung des API 20E-Systems identifiziert worden waren und 10 Referenzstamme wurden unter Verwendung des API 20 EC-Systems und der Assimilationspriifsysteme API 50 CH. API 50 AO und API 50 AA weiter analysiert. Nach numerischer Analyse von 155 ausgewahlten Tests wurden vier Cluster gebildet, die den Spezies Klebsiella pneumoniae subsp. pneumoniae, K. oxytoca, K. planticola und K. terrigena entsprachen. Dann wurden die 26 Tests mit der gro~ten Trennscharfe ermittelt. Es wurde ein Vergleich zwischen der konventionellen Identifizierung mittels API 20 E und den Ergebnissen der numerischen Analyse durchgefiihrt. Das konventionelle Verfahren fiihrte bei 13% der gepriiften Stamme zu falschen Ergebnissen, insbesondere bei den neuen Spezies K. planticola und K. terrigena. Nach numerischer Analyse wurden 17 von 204 Stammen klinischen Ursprungs (8,3%) als K. planticola identifiziert. Nur ein Stamm klinischen Ursprungs wurde als K. terrigena und ein Stamm als K. ornithinolytica identifiziert.

Clinical Strains of Klebsiella

457

Introduction

During the last 10 years, progress has been made in Klebsiella taxonomy. The eighth edition of Bergey's Manual of Determinative Bacteriology recognized only the following three species of Klebsiella: K. pneumoniae, K. rhinoscleromatis and K. ozaenae. In the first edition of Bergey's Manual of Systematic Bacteriology (15), K. rhinoscleromatis and K. ozaenae were considered to be metabolically inactive subspecies of K. pneumoniae, and other species were individualized. According to the DNAJDNA hybridization studies of Jain et al. (13), indole-positive, gelatin-positive strains of K. pneumoniae are now regarded as a new species designated K. oxytoca. Two new species, K. terrigena (12) and K. planticola (syn. K. trevisanii) (1, 6, 8), primarily isolated from environmental samples, have also been described. More recently, Sakazaki et al. (17) proposed the name K. ornithinolytica for ornithine decarboxylase (ODC)-positive, indole-positive strains of Klebsiella formerly known as Klebsiella group 47 (5). This new species is considered to be a distinct biogroup of K. planticola. Although this progress has been achieved, identification of Klebsiella strains remains difficult in the clinical microbiology laboratory. Only 5 species, subspecies or biogroups: K. pneumoniae subsp. pneumoniae, K. pneumoniae subsp. rhinoscleromatis, K. pneumoniae subsp. ozaenae, K. oxytoca 1 and 2, are listed in the API 20 E system's data base and few other tests that can be helpful in the identification of Klebsiella species, are available in commercial systems. Environmental Klebsiella species can be isolated from clinical specimens and cases of nosocomial infection including septicemia caused by K. planticola have been reported (7). It has been the aim of this study firstly, to compare con~entional identification of Klebsiella species using the API 20 E system to numerical analysis of 155 tests, including 147 carbon substrate assimilation tests (API 50 CH, API 50 AO, and API 50 AA systems), and secondly, to evaluate the distribution of clinical strains among the different Klebsiella species.

Material and Methods

Origin and identification of clinical strains. Clinical isolates of Gram-negative bacilli from cardiological and neurological hospitals (Lyon, France) were purified on bromocresol purple agar before initial identification using the API 20 E system (API System, MontalieuVercieu, France). Motility was studied with wet mounts following 24 h peptone-water culture at 3rC (pH = 7.2 ± 0.2). A total of 204 nonmotile strains with the typical profiles of Klebsiella pneumoniae subsp. pneumoniae, K. oxytoca, or unidentified profiles showing Klebsiella-like colonies on agar plates, were collected randomly during a period of 9 months. The strains had been isolated from various clinical samples: urine [101], respiratory tract [100], wounds [56], blood [28], intravascular catheters [12], cerebrospinal fluid [5], female genital tract [5], drainage tubes [4], various liquids [2], ear [11, and stools 11J. Only one strain per patient was included in this study. The strains were further identified using the APr 20 EC system for the identification of coliform bacteria from the environment (API System) (16), which makes it possible to determine lysine decarboxylase (LDC), arginine dihydrolase (ADH), ornithine decarboxylase (ODC), urease, beta-xylosidase, esculin hydrolysis, mucate, glycerol, sorbitol, sucrose, salicin, rhamnose, dulcitol, melezitose, melibiose, inositol, adonitol and sorbose fermentations, Voges-Proskauer test, and indole production after incubation at 30°C for 24 h.

458

D. Monnet,

J. Freney,

Y. Brun, J.-M. Boeufgras, and

J. Fleurette

They were also tested for carbon substrate assimilation using the API 50 CH, API 50 AO, and API 50 AA assimilation systems (API System) which permit the study of 49 carbohydrates, 49 organic acids, and 49 amino acids and amines as sole carbon sources (9). The inoculum was obtained by adding 2 ml of bacterial suspension (McFarland standard 0.5) to 60 ml of the assimilation medium. The isolates were incubated at 30 DC and growth was observed after 1, 2, and 4 days. A reaction was recorded as positive when bacterial development was detected in the cupula. Reference strains. The following 10 reference strains: Klebsiella pneumoniae subsp. pneumoniae ATCC 13882 (American Type Culture Collection, Rockville, Md); K. oxytoca ATCC 13182 and CUETM 77.46 (Collection de l'Unite d'Ecotoxicologie Microbienne, Villeneuve d'Ascq, France); K. planticola ATCC 33531 and ATCC 33558; K. terrigena ATCC 33257, ATCC 33628, ATCC 33629, ATCC 33630 and ATCC 33631; were tested using the API 20 E, the API 20 EC, and the API 50 CH, API 50 AO, and API 50 AA systems. Analysis of data. Eight tests of the API 20 EC system: LDC, ADH, ODC, urease, betaxylosidase, mucate fermentation, Voges-Proskauer test, indole production, and the 147 assimilation tests were analysed numerically on a Microvax II computer (Digital Equipment). The inter-strain distances were calculated using Gower's similarity coefficient (11) and the clusters were formed using the Unweighted Pair Group Method with Average (18). The Descamps method of information gain (INF) (3) was used to establish the diagnosis ability coefficients.

Results Results of conventional identification of the 214 strains using the API 20 E system were as follows: - K. pneumoniae subsp. pneumoniae, 129 strains including the reference strain of this species, the 5 reference strains of K. terrigena, and the 2 reference strains of K. planticola; - K. oxytoca, coded as K. oxytoca 1, 81 strains including the 2 reference strains of this species; - ODC-positive, indole-positive Klebsiella, coded as K. oxytoca 2, 2 strains; - no identification, 2 strains. As a result of numerical analysis of the 8 selected tests from the API 20 EC system and the 147 assimilation tests, 4 clusters, from A to D, were determined on the dendrogram (Fig. 1). Cluster A contained 6 strains, including the 5 reference strains of K. terrigena. Cluster B contained 20 strains, including the 2 reference strains of K. planticola. Cluster C contained 115 strains, including the reference strain of K. pneumoniae subsp. pneumoniae. Cluster D contained 71 strains, including the 2 reference strains of K. oxytoca. The percentages of positive strains in each cluster for the most discriminating tests (25 assimilation tests plus indole production) are given in Table 1. The 2 ODC-positive and indole-positive Klebsiella in this study (strains 213 and 214) were not close to each other on the dendrogram. Strain 213 was included in cluster B. Its characters, from the number of those listed in Table 1 were: assimilation of histamine, DL-kynurenine, DL-3-hydroxybutyrate, Lproline, L-arginine, 5 -ketogluconate, D-turanose, starch, amygdaline, benzoate, Ltryptophan, L-sorbose, and L-ornithine; there was no assimilation of ethanolamine, melezitose, m-hydroxybenzoate, aconitate, L-arabitol, L-tartrate, DL-4-aminobutyrate, dulcitol, D-tagatose, D-arabinose, DL-5-aminovalerate and L-threonine.

Clinical Strains of Klebsiella

459

1. Similarity ')0

100

0 ••

..

"~. ~

"00

~.,

"~.

~~.

"~~

"'" "00 ~o.

-0. "O~

"'" O"~

Ui

~15~

iii-

D + r ..

~:~

0 ••

0 ..

"0" "'~

~~~

gO~: o.

_0. -0' ON

00" 000

0.' ~'o ~".

-.. -,........ ... ... ..",...,

,~o

"~. """

"., ".~ "'0 'o~

".,

o.~

0"' "",. "' 0 • .~

"" 0.' ".0 "O~

UI g~~

§!~

., §OS:

.. I~~ ,. i '· i~e

g "" .N

0'. 0"0 ~,

c

i==: o;~

0'" 0".

i:: g ••

..".

g:::~

8 ~;

"00

0'-'0 o~"

!IE 2"· g 2~

00 0.' 0 ••

g'.

Fig. 1. Phenotypic dendrogram of hierarchical agregation clustering of the 214 Klebsiella strains, cluster A: K. terrigena, cluster B: K. planticola, cluster C: K. pneumoniae subsp. pneumoniae, cluster D: K. oxytoca. Symbols: A, ODC-positive indole-positive Klebsiella; ATCC (American Type Culture Collection, Rockville, Md) strain; +, CUETM (Collection de l'Unite d'Ecotoxicologie Microbienne, Villeneuve d' Ascq, France) strain.

a:; I'"

z··

III B

A

460

D. Monnet, J. Freney, Y. Brun, J.-M. Boeufgras, and J. Fleurette

Table 1. Discriminating carbon substrates and indole production of 214 Klebsiella strains (percentage of positive strains in each cluster) A

B

C

K. terrigena

K. planticola

K. pneumoniae K. subsp. oxytoca pneumoniae

Assimilation of: Ethanolamine Histamine D-melezitose m-hydroxybenzoate DL-kynurenine Indole production

0 83 100 83 0 0

0 100 0 0 95 60

93 0 1 2 0 0

94 0 72 94 0 97

Assimilation of: DL-3-hydroxybutyrate L-proline Aconitate L-arabitol L-tartrate L-arginine DL-4-aminobutyrate Dulcitol D-tagatose 5-ketogluconate D-arabinose D-turanose DL-5 -aminovalerate Starch Amygdaline Benzoate L-tryptophane L-sorbose L-threonine L-ornithine

83 16 83 0 0 66 0 0 0 100 16 0 0 33 16 33 0 100 66 16

100 100 0 0 40 80 65 0 0 100 25 30 0 90 90 100 45 90 10 30

97 100 94 4 72 93 86 28 25 33 68 85 59 99 84 77 0 54 20 46

3 97 86 83 96 1 79 76 77 100 100 51 0 96 82 59 10 99

Cluster and corresponding species

D

11

1

Strain 214 was included in cluster D. Its main characters were: assimilation of ethanolamine, melezitose, m-hydroxybenzoate, L-proline, L-arabitol, L-tartrate, dulcitol, D-tagatose, 5-ketogluconate, D-arabinose, D-turanose, starch, amygdaline, benzoate, and L-sorbose; there was no assimilation of histamine, DL-kynurenine, DL-3hydroxybutyrate, aconitate, L-arginine, DL-4-aminobutyrate, DL-5-aminovalerate, Ltryptophan, L-threonine and L-ornithine. The following 2 strains were not included in any of the clusters: - Strain 115, identified as K. oxytoca by the conventional method; - Strain 161, whose main characters were assimilation of histamine, m-hydroxybenzoate, DL-kynurenine, DL-3-hydroxybutyrate, L-proline, L-arabitol, L-arginine, DL-4aminobutyrate, dulcitol, D-tagatose, 5-ketogluconate, DL-5-aminovalerate, starch, amygdaline, benzoate, L-sorbose, L-threonine, L-ornithine, no assimilation of

461

Clinical Strains of Klebsiella

Table 2. Comparison between conventional identification using the API 20 E system and numerical analysis of 155 tests Conventional identification (API 20 E) Cluster and corresponding species

A

K. pneumoniae subsp. pneumoniae

K.oxytoca

ODC-positive indole-positive Klebsiella

no identification

129 (strains)

81 (strains)

2 (strains)

2 (strains)

5

0

0

1

8

11

1

0

114

0

0

1

2

68

1

0

1

1

0

0

K. terrigena (6 strains) B

K. planticola (20 strains)

C K: pneumoniae subsp. pneumoniae (115 strains)

D K.oxytoca (71 strains)

Not included in any of the clusters (2 strains)

ethanolamine, D-melezitose, aconitate, L-tartrate, D-arabinose, D-turanose, and Ltryptophane, urease-positive, indole-negative, growth at 4°C, no growth at 44.5 0c. Comparison between conventional identification using the API 20 E system and numerical analysis of the 155 tests has been reproduced in Table 2. Discussion Numerical analysis of 155 tests provides accurate identification of the 4 species K. pneumoniae subsp. pneumoniae, K. oxytoca, K. planticola and K. terrigena. The percentages of positivity of the assimilation tests were, for the common tests, almost identical with those of Gavini et aI. (to) and Naemura et al. (14). Strains of cluster A were comparable to the environmental strains of group L of Gavini et al. and of group 2 of Naemura et aI., and corresponded to the description of K. terrigena (12).

462

D. Monnet, ]. Freney, Y. Brun, ].-M. Boeufgras, and J. Fleurette

Strains of cluster B were comparable to the environmental strains of group K of Gavini et al. and to histamine-positive strains of group 4 of Naemura et al., and they corresponded to the description of K. planticola (1, 6). Strains of clusters C and D were comparable to those of groups J and M of Gavini et al. and of groups 3 and 1 of Naemura et al. respectively. They corresponded to the descriptions of K. pneumoniae subsp. pneumoniae and K. oxytoca (15) respectively. Strain 213 was included in cluster B and corresponded to the description of K. ornithinolytica (17). Strain 214 was included in cluster D and did not correspond to the description of K. ornithinolytica because its fermentation tests were positive for both melezitose and dulcitol. This strain can be considered as an atypical ODC-positive indole-positive Klebsiella. Confirmation of these 2 identifications should be obtained by DNA-DNA hyrn-idization studies. Only 2 out of 214 strains (0.9%) could not be included in any of the clusters and were finally identified using an experimental assimilation gallery which permits the study of 100 carbon substrates (2). Strain 115 was initially identified as K. oxytoca using the API 20 E system but could not be included in cluster D because of the clustering level. Its identification as K. oxytoca was confirmed with the aid of the experimental assimilation gallery. Strain 161 was finally identified as atypical Klebsiella when applying this assimilation gallery. Comparison of the 2 methods shows that the conventional method (API 20 E system) generally leads to accurate identification of the following 3 species: K. pneumoniae subsp. pneumoniae, K. oxytoca and K. ornithinolytica. Only 2 strains initially identified as K. pneumoniae subsp. pneumoniae using the API 20 E system were furthermore included in cluster D and can be considered as indole-negative strains of K. oxytoca. Two other strains, 1 initially identified as K. pneumoniae subsp. pneumoniae and 1 as K. oxytoca, were not included in any of the clusters. Their final identification with the aid of the experimental assimilation gallery has been already discussed. However, the conventional method will always lead to incorrect identification of the 2 new species, K. terrigena and K. planticola, as they are not included in the API 20 E system's data base. Of the 6 strains identified as K. terrigena after numerical analysis (cluster A), 5 were initially identified as K. pneumoniae subsp. pneumoniae and 1 remained unidentified by the API 20 E system. Of the 19 strains identified as K. planticola after numerical analysis (cluster B), 8 were initially identified as K. pneumoniae subsp. pneumoniae and 11 as K. oxytoca using the API 20 E system. According to the results of the numerical analysis, only 1 of the 2 strains identified as ODC-positive indole-positive Klebsiella with the API 20 E system was finally identified as K. ornithinolytica. In conclusion, the conventional method resulted in incorrect identification of 13% of the strains tested. Seventeen out of 204 strains of clinical origin (8.3 %) were identified as K. planticola after numerical analysis. Confirmation of these identifications should be obtained by DNA-DNA hybridization studies. Of these strains, 10 had been isolated from the respiratory tract, S from urine, 1 from a wound and 1 from a drainage tube. Nosocomial infections by K. planticola have already been described (7) and this study shows that the recovery of K. planticola and K. terrigena in clinical specimens is by no means negligible. Accurate identification of these species can be obtained when applying, after the

Clinical Strains of Klebsiella

463

conventional API 20 E system, with supplementary tests such as growth at 44.5°C, lOOC and 4°C, and pectate liquefaction (4, 15) but growth at low temperatures is generally slow and culture media should sometimes be observed for 2 weeks. The pectate medium is also difficult to prepare. An alternative could consist in 1 or 2 selected carbon substrate assimilation tests such as histamine, ethanolamine, melezitose, m-hydroxybenzoate, DL-kynurenine, but these substrates are not commercially available as individual tests and cannot be easily tested. Acknowledgments. We would like to thank P. A. D. Grimont and E. Ageron for final identification of fastidious strains, F. Gavini for providing the reference strains and F. Allard for her help in data entry and numerical analysis.

References 1. Bagley, S. T., R. J. Seidler, and D. J. Brenner: Klebsiella planticola sp. nov.: a new species of Enterobacteriaceae found primarily in nonclinical environments. Curro Microbiol. 6 (1981) 105-109 2. Bouvet, O. M. M., P. A. D. Grimont, C. Richard, E. Aldova, O. Hausner, and M. Gabrhelova: Budvicia aquatica gen. nov., sp. nov.: a hydrogen sulfide-producing member of the Enterobacteriaceae. Int. J. System. Bact. 35 (1985) 60-64 3. Descamps, P. and M. Veron: Une methode de choix des caracteres d'identification basee sur Ie theoreme de Bayes et la mesure de I'information. Ann. Microbiol. (Inst. Pasteur) 132 B (1981) 157-170 . 4. Edwards, P. R. and W. H. Ewing: Identification of Enterobacteriaceae, 4th ed. Elsevier Science Publishing Co., New York (1986) 5. Farmer, J. J. III, B. R. Davis, F. W. Hickman-Brenner, A. McWhorther, G. P. HuntleyCarter, M. A. Asbury, C. Riddle, H. G. Wathen-Grady, C. Elias, G. R. Fanning, A. G. Steigerwalt, C. M. O'Hara, G. K. Morris, P. B. Smith, and D. J. Brenner: Biochemical identification of new species and biogroups of Enterobacteriaceae isolated from clinical specimens. ]. Clin. Microbiol. 21 (J 985) 46-76 6. Ferragut, c., D. Izard, F. Gavini, K. Kersters, J. De Ley, and H. Leclerc: Klebsiella trevisanii: a new species from water and soil. Int. ]. System. Bact. 33 (1983) 133-142 7. Freney, J., F. Gavini, H. Alexandre, S. Madier, D. Izard, H. Leclerc, and J. Fleurette: Nosocomial infection and colonization by Klebsiella trevisanii. J. Clin. Microbiol. 23 (1986) 948-950 8. Gavini, F., D. Izard, P. A. D. Grimont, A. Beji, E. Ageron, and H. Leclerc: Priority of Klebsiella planticola Bagley, Seidler and Brenner 1982 over Klebsiella trevisanii Ferragut, Izard, Gavini, Kersters, De Ley and Leclerc 1983. Int. J. System. Bact. 36 (1986) 486-488 9. Gavini, F., D. Izard, H. Leclerc, M. Desmonceaux, and J. P. Gayral: Carbon sources assimilation tests: comparison between a conventional method and a microtechnic (API), in study of Enterobacteriaceae. Zbl. Bakt. Hyg., I. Abt. Orig. C 1 (1980) 182-187 10. Gavini, F., H. Leclerc, B. Lefebvre, C. Ferragut, and D. Izard: Etude taxonomique d'Enterobacteries appartenant ou apparentees au genre Klebsiella. Ann. Microbiol. (Inst. Pasteur) 128 B (1977) 45-59 11. Gower, J. c.: A general coefficient of similarity and some of its properties. Biometrics 27 (1971) 857-874 12. Izard, D., C. Ferragut, F. Gavini, K. Kersters, J. De Ley, and H. Leclerc: Klebsiella terrigena, a new species from soil and water. Int. J. System. Bact. 31 (1981) 116-127 13. Jain, K., K. Radsak, and W. Mannheim: Differentiation of the Oxytocum group from Klebsiella by deoxyribonucleic acid-deoxyribonucleic acid hybridization. Int. J. System. Bact. 24 (1974) 402-407

464

D. Monnet, J. Freney, Y. Brun, J.·M. Boeufgras, and J. Fleurette

14. Naemura, L. G., S. T. Bagley, R. J. Seidler, J. B. Kaper, and R. R. Colwell: Numerical taxonomy of Klebsiella pneumoniae strains isolated from clinical and nonclinical sources. Curr. Microbiol. 2 (1979) 175-180 15. Orskov, I.: Genus V Klebsiella Trevisan 1885, 105 AL • In: Bergey's manual of systematic bacteriology, vol. 1, ed. by N. R. Krieg, pp. 461-465. The Williams and Wilkins Co., Baltimore (1984) 16. Renaud, F., J. Freney, J. M. Boeufgras, and J. Fleurette: Comparaison de deux systemes d'identification d'enterobacteries coliformes nouvellement decrites ou rarement rencontrees en clinique. Ann. BioI. Clin. 48 (1990) 111-115 17. Sakazaki, R., K. Tamura, Y. Kosako, and E. Yoshizaki: Klebsiella ornithinolytica sp. nov., formerly known as ornithine-positive Klebsiella oxytoca. Curro Microbiol.18 (1989) 201-206 18. Sneath, P. H. A. and R. R. Sokal: Numerical taxonomy - The principles and practice of numerical classification. W. H. Freeman and Co., San Francisco (1973)

Dr. Dominique Monnet, Laboratoire de Bacteriologie, Hopital de la Croix-Rousse, 93 Grande rue de la Croix-Rousse, 69317 Lyon Cedex 04, France