application ofgas-liquid chromatographic analysis fatty ... · journalofclinical microbiology, feb....

8
Vol. 29, No. 2 JOURNAL OF CLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 0095-1137/91/020315-08$02.00/0 Copyright © 1991, American Society for Microbiology Application of Gas-Liquid Chromatographic Analysis of Cellular Fatty Acids for Species Identification and Typing of Coagulase-Negative Staphylococci PIRKKO KOTILAINEN,l.2* PENTTI HUOVINEN,l AND ERKKI EEROLA' Department of Medical Microbiology, Turku University,' and Department of Medicine, Turku University Central Hospital,2 Turku, Finland Received 21 August 1990/Accepted 7 November 1990 Gas-liquid chromatography (GLC) of bacterial cellular fatty acids was used to analyze 264 isolates of coagulase-negative staphylococci, of which 178 were Staphylococcus epidermidis. The presence and amounts of individual fatty acids were determined to generate fatty acid profiles for each of the seven coagulase-negative species tested. The fatty acid profiles were then analyzed by computerized correlation and cluster analysis to calculate mean correlation values between isolates belonging to the same or different species, as well as to establish cluster analysis dendrograms. These data ultimately allowed the clustering of individual samples into species-specific clusters. Species identification by the GLC clustering was highly consistent with species identification by biochemical assays; the results were similar in 92.4% of the cases. The GLC profile correlation analysis was further used to analyze multiple blood isolates from 60 patients in order to determine the usefulness of this methodology in establishing identity, as well as differences, between consecutive patient isolates. The correlation between those multiple S. epidermidis isolates determined to be identical by standard techniques (such as the antibiogram, biotype, and plasmid profile) was significantly (P < 0.001) higher than that between random isolates of the same species. The correlation coefficient was >97 for 40 (97.6%) of the 41 patients with multiple identical blood isolates, compared with <95 in all 19 (100.0%) patients with multiple nonidentical isolates. The successful use of the computerized GLC analysis in this study demonstrated its appropriate application for species identification and typing of coagulase-negative staphylococci. Coagulase-negative staphylococci are among the most important nosocomial pathogens (2, 13, 16, 24). These mi- croorganisms are generally associated with implanted for- eign bodies and impaired host defenses (15, 18, 34). Staph- ylococcus epidermidis is the coagulase-negative species most prevalent in disease; however, other species may also be pathogenic (3, 20, 25). Therefore, identification of these bacteria to the species level is needed to provide precise etiological diagnoses of coagulase-negative staphylococcal infections. At present, the genus Staphylococcus comprises 27 different species, most of which are coagulase negative (30). In clinical laboratories, identification of the specific coagulase-negative staphylococcal species has so far been based primarily on various biochemical tests (25). In addi- tion to the conventional methods of Kloos and Schleifer, several commercial testing systems are available for this purpose (25). These assays allow species identification with an accuracy of 70 to >90% (1, 5, 6, 11, 12, 25). Recently, alternative methods have been introduced which identify different coagulase-negative staphylococcal species by the chemical compounds present in the bacterial cells. These include electrophoretic profiles of the total bacterial proteins as well as of the penicillin-binding proteins (26, 33). In addition, efforts have been made to apply bacterial cellular fatty acid composition for species identification of staphylo- cocci (7, 21, 30). By gas-liquid chromatographic (GLC) analysis (8, 30), the fatty acid profiles of the isolates belong- ing to the genus Staphylococcus have been shown to be distinct from those of other bacterial species. The fatty acid profiles of different species of coagulase-negative staphylo- * Corresponding author. cocci, on the other hand, proved qualitatively to be highly similar (7, 8, 21, 30). However, quantitative differences in the fatty acid composition between various species were obvious in these studies. Because of the ubiquitous nature of coagulase-negative staphylococci in the environment, effective methods are needed for identification of clinically significant strains. The most common approaches for epidemiological typing include the antimicrobial susceptibility pattern (antibiogram), bio- chemical characterization (biotyping), bacteriophage sus- ceptibility pattern (phage typing), serological typing (sero- typing), and detection of bacterial adherence and slime production by the tube adherence test (4, 22, 23, 25). In addition, molecular techniques, including plasmid profile analysis as well as DNA-DNA hybridization and restriction enzyme analysis of chromosomal and plasmid DNA, have been applied for strain characterization (9, 10, 22, 23, 28). Many of these assays (such as antibiograms, biotyping, the tube adherence test, and molecular analysis of plasmids) are readily available in most clinical laboratories. Others (such as phage typing, serotyping, restriction endonuclease map- ping, and DNA-DNA hybridization) are predominantly re- search tools. A computer-driven GLC analysis method (8) has recently been developed in our laboratory for identification of bacte- rial strains. This is operationally accomplished by analysis of bacterial fatty acid profiles. The purpose of the present study was to analyze the fatty acid composition of different species of coagulase-negative staphylococci. The mean relative amount of each detected fatty acid in the isolates, belonging to seven coagulase-negative species, was examined to test whether the bacterial fatty acid profiles could be used for reliable speciation of coagulase-negative staphylococci. In 315 on December 15, 2020 by guest http://jcm.asm.org/ Downloaded from

Upload: others

Post on 25-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

Vol. 29, No. 2JOURNAL OF CLINICAL MICROBIOLOGY, Feb. 1991, p. 315-3220095-1137/91/020315-08$02.00/0Copyright © 1991, American Society for Microbiology

Application of Gas-Liquid Chromatographic Analysis of CellularFatty Acids for Species Identification and Typing of

Coagulase-Negative StaphylococciPIRKKO KOTILAINEN,l.2* PENTTI HUOVINEN,l AND ERKKI EEROLA'

Department of Medical Microbiology, Turku University,' and Department of Medicine, Turku UniversityCentral Hospital,2 Turku, Finland

Received 21 August 1990/Accepted 7 November 1990

Gas-liquid chromatography (GLC) of bacterial cellular fatty acids was used to analyze 264 isolates ofcoagulase-negative staphylococci, of which 178 were Staphylococcus epidermidis. The presence and amounts ofindividual fatty acids were determined to generate fatty acid profiles for each of the seven coagulase-negativespecies tested. The fatty acid profiles were then analyzed by computerized correlation and cluster analysis tocalculate mean correlation values between isolates belonging to the same or different species, as well as toestablish cluster analysis dendrograms. These data ultimately allowed the clustering of individual samples intospecies-specific clusters. Species identification by the GLC clustering was highly consistent with speciesidentification by biochemical assays; the results were similar in 92.4% of the cases. The GLC profile correlationanalysis was further used to analyze multiple blood isolates from 60 patients in order to determine theusefulness of this methodology in establishing identity, as well as differences, between consecutive patientisolates. The correlation between those multiple S. epidermidis isolates determined to be identical by standardtechniques (such as the antibiogram, biotype, and plasmid profile) was significantly (P < 0.001) higher thanthat between random isolates of the same species. The correlation coefficient was >97 for 40 (97.6%) of the 41patients with multiple identical blood isolates, compared with <95 in all 19 (100.0%) patients with multiplenonidentical isolates. The successful use of the computerized GLC analysis in this study demonstrated itsappropriate application for species identification and typing of coagulase-negative staphylococci.

Coagulase-negative staphylococci are among the mostimportant nosocomial pathogens (2, 13, 16, 24). These mi-croorganisms are generally associated with implanted for-eign bodies and impaired host defenses (15, 18, 34). Staph-ylococcus epidermidis is the coagulase-negative speciesmost prevalent in disease; however, other species may alsobe pathogenic (3, 20, 25). Therefore, identification of thesebacteria to the species level is needed to provide preciseetiological diagnoses of coagulase-negative staphylococcalinfections. At present, the genus Staphylococcus comprises27 different species, most of which are coagulase negative(30). In clinical laboratories, identification of the specificcoagulase-negative staphylococcal species has so far beenbased primarily on various biochemical tests (25). In addi-tion to the conventional methods of Kloos and Schleifer,several commercial testing systems are available for thispurpose (25). These assays allow species identification withan accuracy of 70 to >90% (1, 5, 6, 11, 12, 25). Recently,alternative methods have been introduced which identifydifferent coagulase-negative staphylococcal species by thechemical compounds present in the bacterial cells. Theseinclude electrophoretic profiles of the total bacterial proteinsas well as of the penicillin-binding proteins (26, 33). Inaddition, efforts have been made to apply bacterial cellularfatty acid composition for species identification of staphylo-cocci (7, 21, 30). By gas-liquid chromatographic (GLC)analysis (8, 30), the fatty acid profiles of the isolates belong-ing to the genus Staphylococcus have been shown to bedistinct from those of other bacterial species. The fatty acidprofiles of different species of coagulase-negative staphylo-

* Corresponding author.

cocci, on the other hand, proved qualitatively to be highlysimilar (7, 8, 21, 30). However, quantitative differences inthe fatty acid composition between various species wereobvious in these studies.Because of the ubiquitous nature of coagulase-negative

staphylococci in the environment, effective methods areneeded for identification of clinically significant strains. Themost common approaches for epidemiological typing includethe antimicrobial susceptibility pattern (antibiogram), bio-chemical characterization (biotyping), bacteriophage sus-ceptibility pattern (phage typing), serological typing (sero-typing), and detection of bacterial adherence and slimeproduction by the tube adherence test (4, 22, 23, 25). Inaddition, molecular techniques, including plasmid profileanalysis as well as DNA-DNA hybridization and restrictionenzyme analysis of chromosomal and plasmid DNA, havebeen applied for strain characterization (9, 10, 22, 23, 28).Many of these assays (such as antibiograms, biotyping, thetube adherence test, and molecular analysis of plasmids) arereadily available in most clinical laboratories. Others (suchas phage typing, serotyping, restriction endonuclease map-ping, and DNA-DNA hybridization) are predominantly re-search tools.A computer-driven GLC analysis method (8) has recently

been developed in our laboratory for identification of bacte-rial strains. This is operationally accomplished by analysis ofbacterial fatty acid profiles. The purpose of the present studywas to analyze the fatty acid composition of different speciesof coagulase-negative staphylococci. The mean relativeamount of each detected fatty acid in the isolates, belongingto seven coagulase-negative species, was examined to testwhether the bacterial fatty acid profiles could be used forreliable speciation of coagulase-negative staphylococci. In

315

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Page 2: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

316 KOTILAINEN ET AL.

addition, we tested the adequacy of this technique as a novelscheme for the typing of coagulase-negative staphylococci.To do so, we compared the degree of GLC fatty acid profilesimilarity in consecutive blood isolates previously desig-nated as identical or nonidentical by conventional typingmethods.

MATERIALS AND METHODS

Bacterial strains and identification. Four reference strainswere obtained from the American Type Culture Collection(Rockville, Md.), including S. epidermidis ATCC 35983(RP12), ATCC 35984 (RP62A), and ATCC 14990 and S.hominis ATCC 35981 (RP14). Three additional referencestrains (S. epidermidis KH11, KH6, and V2) were providedby Georg Peters from the Institute of Hygiene, University ofCologne, Cologne, West Germany. A total of 312 coagulase-negative staphylococcal isolates were recovered from bloodcultures at Turku University Central Hospital and TurkuCity Hospital during the years 1983 to 1989. These included135 isolates from 60 patients who had more than one positiveblood culture and 177 isolates from patients with singlepositive blood cultures. Stock cultures of the strains weremaintained at -70°C in tryptone soy broth (Oxoid Ltd.,Basingstoke, England) containing 20% glycerol.

Staphylococci were identified by standard methods, in-cluding Gram staining as well as catalase and tube coagulasetests. All isolates were speciated by the API Staph-Tracsystem (Analytab Products, Plainview, N.Y.). The isolatesnot groupable by this system were identified to the specieslevel by a new test system, the ATB 32 STAPH (APISystem, SA, La Balme-les-Grottes, Montalieu-Vercieu,France). This system contains 26 biochemical reactions (26).

Identity between the multiple staphylococcal isolates re-covered from a given patient was determined by using anantibiotic susceptibility profile, by biotyping using the bio-chemical assays described above, by positivity or negativityin the tube adherence test, and by using plasmid profileanalysis. Plasmid profiles were determined as described byEtienne et al. (9). When multiple isolates from the samepatient were determined to be identical by these techniques,they were defined, for the purposes of this study, to be thesame strain.GLC analysis. The GLC analysis of the bacterial cellular

fatty acids was performed as described previously (8, 19).The bacteria were cultured for GLC on Fastidious anaerobeagar (LAB M, Bury, England) for 18 h at 37°C, collectedfrom the plates, saponified, methylated, and analyzed asdescribed earlier (8). In brief, the collected bacteria wereincubated for 30 min at 100°C in 15% (wt/vol) NaOH in 50%aqueous methanol, and they were acidified to pH 2 with 6 Naqueous HCl in CH30H. The methylated fatty acids werethen extracted with ethyl ether and hexane. The GLCanalysis was performed with an HP5890A gas chromato-graph (Hewlett-Packard) and an Ultra 2, 004-11-09B fused-silica capillary column (0.2 mm by 25 m; cross-linked 5%phenylmethyl silicone; Hewlett-Packard). Ultra-high-purityhelium was used as the carrier gas. The GLC settings wereas follows: injection port temperature, 250°C; detector tem-perature, 300°C; initial column temperature, 170°C (increas-ing at 5°C per min up to 270°C at 20 min); total analysis time,25 min; sample volume, 1 ,u. The peak retention time andpeak area values were recorded by an HP3392A integrator(Hewlett-Packard).The individual fatty acids were identified by comparing

their retention times to those of a bacterial fatty acid

standard (bacterial acid methyl ester mix CP, 4-7080; Su-pelco, Inc., Bellefonte, Pa.). The relative peak areas of thesefatty acids in each isolate were calculated. Also, the meanrelative peak areas of each of the fatty acids within eachcoagulase-negative staphylococcal species were calculated.

Processing of the GLC data. The GLC results, includingthe retention time values and peak areas of all detectedpeaks in the chromatograph, were transferred to a computer.Correlation and cluster analyses of the data were performedas described earlier (8, 31, 32).

All GLC profiles were compared as pairs to calculatesimilarity indices between individual sample pairs. Thesevalues were further subjected to cluster analysis and pre-sented as dendrograms in order to determine the capacity,efficiency, and reliability of the GLC methodology in group-ing staphylococcal isolates into species-specific clusters. Inaddition, mean values and standard deviations of the simi-larity indices within each species and between differentspecies groups were calculated.

Statistical analysis. Differences in the amounts of the fattyacids between different coagulase-negative species wereevaluated statistically by using the Student t test. This testwas also used to analyze differences in the mean correlationvalues between multiple identical and nonidentical patientisolates.

RESULTS

Bacterial strains. Coagulase-negative staphylococcalblood isolates were grouped by the API Staph-Trac and ATB32 STAPH procedures into seven different species: S. epi-dermidis, S. warneri, S. hominis, S. haemolyticus, S. capi-tis, S. simulans, and S. lugdunensis. All 177 single bloodisolates were included in the mean fatty acid compositionanalysis of these seven coagulase-negative species. Charac-terization of the 135 multiple isolates from 60 patientsrevealed a subset of 96 isolates from 41 patients. In this lattergroup, two to four isolates from each patient were found tobe identical by the antibiogram, biotype, tube adherence testresult, and plasmid profile. Therefore, these two to fourisolates were considered to be the same strain for eachpatient. The duplicate isolates were thus excluded from thefatty acid analysis. The remaining 39 multiple isolates from19 patients proved to be nonidentical. From nine patientswere recovered isolates belonging to different species. Fromthe 10 remaining patients were recovered isolates of thesame species, but with a different antibiotic susceptibilitypattern and plasmid profile. All these isolates were includedin the fatty acid analysis. Thus, the final strain collectionconsisted of 264 isolates of coagulase-negative staphylo-cocci. These included the 177 single blood isolates, 80 of themultiple isolates (after exclusion of the multiple identicalisolates), and finally, 7 reference strains. Of these isolates,178 were grouped as S. epidermidis, 59 were grouped as S.warneri, 10 were grouped as S. hominis, 7 were grouped asS. haemolyticus, 6 were grouped as S. capitis, 2 weregrouped as S. simulans, and 2 were grouped as S. lugdun-ensis.

Bacterial fatty acid composition. The mean fatty acidcompositions of the isolates of the seven coagulase-negativespecies studied are shown in Fig. 1, and the fatty acidcompositions of five of these species, each represented bymore than two isolates, are shown in Table 1. A total of 21different fatty acids were detected in the chromatograms. Ofthese 21 fatty acids, 16 were identified by the fatty acidstandard. The designations of the identified fatty acids are

J. CLIN. MICROBIOL.

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Page 3: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

CHARACTERIZATION OF COAGULASE-NEGATIVE STAPHYLOCOCCI 317

50r S.epidermidis403020

5040

30

20

10

50

40

30

20

10

sc

4C

3020

lCI

A' ... id ...1 'b i 33 03 405

- S hominis

10 155' 10.lb. 15 20 25 30 3'A "0 45

S capitis

50 -

40 -

30 -

20-10

50-40 -

30 -

20-

10

50 -

40 -

30-

20-10-

i i5-P. 25 3'035I 4bpo

45

) S.lugdunensis

A- I_it _ "^

rwr|| wggs lnss.1 1

S. worneri

n

S. haemolyticus

S. simulans

5

I'fh..Al

I _

n10'l 'iS' 20 ~ 3 3- 'Zb-4

5 10 15 20 25 30 35 40 45FIG. 1. Cellular fatty acid composition in seven species of coagulase-negative staphylococci. Mean relative peak areas ± standard

deviation of the fatty acids are indicated on the vertical axes as percentages. The samples were analyzed for the presence of 46 fatty acids;numbers on the horizontal axis refer to the fatty acid elution order from the GLC column. The 21 fatty acids detected in the samples are listedin Table 1 under the same elution indicator numbers. Those 16 fatty acids identified by using the fatty acid standard are designated in Table 2.

given in Table 2. Essentially the same fatty acids werepresent in all analyzed coagulase-negative species. In gen-eral, the prevalence of any given fatty acid in the specimenswas 100% if the relative amount of that fatty acid was greaterthan 1% of the total cellular fatty acid. Differences in theapparent prevalence were observed only in those fatty acidswith the mean amount below 1% of the total content.The most abundant fatty acids of the staphylococci were

as follows: Ci.15:0, Ca15:0, Ci-17:0, Ca-17:0, and C18:0. Therelative amounts of the fatty acids Cji160, C16:0, C18:2912 ,C18:19, and C20:0 and of three uncharacterized fatty acids,defined herein by the elution numbers 9, 14, and 35, weremoderate or small. The relative amounts of the remainingeight fatty acids were insignificant.Only the S. epidermidis, S. warneri, S. hominis, S. hae-

molyticus, and S. capitis species (each represented by morethan two isolates) were analyzed statistically. Statisticaldifferences were calculated for those fatty acids with a meanamount of >1% of the total fatty acid content and a preva-lence of 100%. Only the major differences in the predomi-nant fatty acids are considered herein.Compared with the other species, S. epidermidis dis-

played a significantly larger (P < 0.010) amount of the fattyacid C18:0 and a smaller (P < 0.001) amount of the fatty acidCa15:0. The amount of this same fatty acid Ca 15:0 wassignificantly (P < 0.001) larger in S. warneri, which incontrast contained a significantly (P < 0.002) smaller amountof C18:0. In S. hominis, the amount of the fatty acid Cai17:0was significantly (P < 0.001) smaller than in the four otherspecies; S. hominis contained a significantly (P < 0.050)larger amount of the fatty acid C20:0. The amount of the fattyacid Ca17:0 was significantly (P < 0.020) larger in S. hae-molyticus than in the other species. Finally, the S. capitisspecies displayed a significantly (P < 0.001) larger amount ofthe fatty acid Ci 17:0 than did the others. There was consid-erable overlapping in the statistical differences betweenvarious species. For instance, the quantities of anotheruncharacterized fatty acid, defined by the elution number 14,were significantly (P < 0.001) larger in both S. epidermidisand S. hominis than in the other three species, the quantitiesof Ci 15:0 were significantly (P < 0.01) smaller in both S.warneri and S. haemolyticus than in the others, and thequantities of Cji17:0 were significantly (P < 0.020) smaller inboth S. hominis and S. haemolyticus than in the others.

I

VOL. 29, 1991

N.. .... I ..16 i5 0

t 10 15 4

-n ni

Pk A .

0

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Page 4: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

318 KOTILAINEN ET AL.

TABLE 1. Mean composition of cellular fatty acids in five species of coagulase-negative staphylococcia

S. epidermidis S. warneri S. hominis S. haemolyticus S. capitisNo.b Fatty acid (n = 178) (n = 59) (n = 10) (n = 7) (n = 6)

designationcPrev Peak area Prev Peak area Prev Peak area Prev Peak area Prev Peak area

8 C12:0 0.6 0.00 ± 0.04 0.0 0.0 0.0 0.09 Cn 99.4 1.23 ± 0.32 100.0 1.35 ± 0.31 100.0 2.05 ± 0.42 100.0 1.55 ± 0.44 100.0 1.05 ± 0.3014 Cn 100.0 1.86 ± 0.34 100.0 0.82 ± 0.21 100.0 2.06 ± 0.72 100.0 0.68 ± 0.23 100.0 0.60 ± 0.1916 C14:0 100.0 0.61 ± 0.22 53.4 0.13 ± 0.14 100.0 0.49 ± 0.21 57.1 0.17 ± 0.16 66.7 0.17 ± 0.1618 Ci15:0 100.0 7.37 ± 1.17 100.0 5.11 ± 0.95 100.0 8.23 ± 1.07 100.0 5.56 ± 1.24 100.0 7.76 ± 0.8619 Ca-15:0 100.0 40.59 ± 2.39 100.0 48.27 ± 1.70 100.0 45.50 + 2.51 100.0 45.41 + 1.36 100.0 45.48 ± 1.3220 Cn 2.9 0.00 ± 0.03 1.7 0.00 ± 0.02 0.0 0.0 0.021 C15:0 55.2 0.20 ± 0.22 50.0 0.14 ± 0.16 10.0 0.03 ± 0.08 42.9 0.17 + 0.22 16.7 0.08 ± 0.1724 Cj16:0 100.0 2.06 ± 0.33 100.0 1.20 ± 0.22 100.0 1.47 ± 0.39 100.0 0.85 ± 0.28 100.0 0.72 ± 0.1825 C16:19 11.6 0.03 ± 0.07 1.7 0.00 ± 0.02 0.0 0.0 0.026 C16:0 100.0 2.88 ± 0.59 100.0 2.10 ± 0.44 100.0 1.91 ± 0.32 100.0 2.15 ± 0.44 100.0 1.73 ± 0.5128 Ci17:0 100.0 6.49 ± 1.00 100.0 6.45 ± 1.01 100.0 5.63 ± 0.73 100.0 5.47 ± 0.99 100.0 7.96 ± 1.0029 Ca17:0 100.0 15.76 ± 2.31 100.0 22.03 ± 2.14 100.0 13.12 + 2.93 100.0 24.09 ± 1.74 100.0 19.62 ± 2.3631 C17:0 97.1 0.36 ± 0.12 94.8 0.35 ± 0.18 50.0 0.16 ± 0.16 85.7 0.28 ± 0.15 66.7 0.19 ± 0.1935 Cn 100.0 1.63 ± 0.32 100.0 0.75 ± 0.20 100.0 1.93 ± 0.68 100.0 0.53 ± 0.14 100.0 0.57 ± 0.1136 C18:29'12 100.0 1.56 ± 0.69 100.0 2.01 ± 0.56 100.0 1.04 ± 0.08 100.0 1.04 ± 0.22 100.0 2.26 ± 0.5337 C18:19 100.0 1.61 ± 0.62 91.4 0.95 ± 0.43 70.0 0.82 ± 0.70 71.4 0.60 ± 0.42 100.0 1.30 ± 0.3139 C18:0 100.0 9.98 ± 1.67 100.0 5.29 ± 1.13 100.0 8.52 ± 1.20 100.0 8.28 ± 1.23 100.0 6.90 ± 1.1544 C19:0 99.4 0.61 ± 0.20 84.5 0.25 ± 0.16 100.0 0.68 ± 0.19 100.0 0.32 ± 0.10 83.3 0.20 ± 0.1445 Cn 95.3 0.82 ± 0.44 98.3 0.99 ± 0.40 100.0 1.09 ± 0.38 85.7 0.60 ± 0.42 83.3 0.90 ± 0.4746 C20:0 100.0 4.35 ± 1.34 100.0 1.81 ± 0.77 100.0 5.28 ± 1.55 100.0 2.23 ± 0.44 100.0 2.52 ± 0.26

a Prevalence (Prev) indicates the percentage of isolates with detectable fatty acid within each species. Peak area is expressed as the mean + standard deviation.b The samples were analyzed for the presence of 46 fatty acids; No. refers to the elution order from the GLC column. These indicator numbers are used to define

the detected fatty acids in the text, in Fig. 1 (horizontal axes), and in Table 2.c Of the 46 analyzed fatty acids, only the 21 fatty acids detected in the samples are listed. C. indicates an uncharacterized fatty acid. The fatty acids, identified

according to the fatty acid standard, are designated in Table 2.

Furthermore, all five coagulase-negative species differedsignificantly from each other in the quantities of the fattyacid Ca17:0 detected.

Correlation analysis. Similarity indices were calculated by

TABLE 2. Fatty acids isolated from coagulase-negative staphylococcia

No.b Designationc Name

8 C12:0 Dodecanoate16 C14:0 Tetradecanoate18 Cji15:0 13-Methyltetradecanoate19 Ca-15:0 12-Methyltetradecanoate21 C15:0 Pentadecanoate24 Cji16:0 14-Methylpentadecanoate25 C16:19 cis-9-Hexadecenoate26 C16:0 Hexadecanoate28 Cji17:0 15-Methylhexadecanoate29 Ca17:0 14-Methylhexadecanoate31 C17:0 Heptadecanoate36 C18:29'12 cis-9,12-Octadecadienoate37 C18:19 cis-9-Octadecenoate39 C18:0 Octadecanoate44 C19:0 Nonadecanoate46 C20:0 Eicosanoatea Staphylococci were analyzed for the presence of 46 fatty acids. Listed are

the 16 fatty acids detected in the test samples and identified according to thefatty acid standard.

b Numbering refers to the elution order from the GLC column. Theseindicator numbers are used to define the detected fatty acids in the text, in Fig.1 (horizontal axes), and in Table 1.

c The number before the colon refers to the number of carbon atoms, thenumber after the colon refers to the number of double bonds, i indicates abranched-chain acid with the branched methyl group at the iso position, anda indicates a branched-chain acid with the branched methyl group at theanteiso position.

comparing the isolates within the same species as pairs andalso by comparing the isolates of one species to the isolatesof a different species. The means of similarity indices werethen calculated within each species and between differentspecies. Isolates belonging to the same species resembledeach other more closely than isolates belonging to differentspecies: the mean correlation of the fatty acid profiles of theisolates of the same species varied from 85.54 +7.68 in S.hominis to 95.65 in S. simulans and the mean correlation ofthe isolates of different species varied from 62.28 ± 5.87between S. haemolyticus and S. simulans to 86.76 ± 4.56between S. haemolyticus and S. warneri (Table 3). The GLCfatty acid profile similarities between the seven coagulase-negative species studied are presented in Fig. 2 as a dendro-gram after cluster analysis of the mean correlations betweenthese species. S. epidermidis and S. hominis formed oneloosely related cluster, S. warneri, S. haemolyticus, S.capitis, and S. Iugdunensis formed another, and S. simulanswas grouped in a separate cluster by itself.The individual isolates were also analyzed by a weighted-

pair cluster analysis. Results are presented as a dendrogramin Fig. 3. Isolates belonging to the same species formedseparate clusters. Of the 178 S. epidermidis isolates, 168formed a defined cluster. Also included in this cluster werefive S. hominis and two S. warneri isolates. Another clusterwas formed by 55 of the 59 S. warneri, 1 S. haemolyticus,and 3 S. epidermidis isolates. Additionally, a subcluster ofsix of the seven S. haemolyticus isolates was grouped intothis cluster. Still other distinct clusters were formed by 6 S.capitis, by 2 S. lugdunensis, by 2 S. simulans, and by 5 of the10 S. hominis isolates. Finally, seven S. epidermidis and twoS. warneri isolates were adjusted to fall outside all clusters.When the species identification of staphylococci by the

biochemical assays (API Staph-Trac and ATB 32 STAPH)

J. CLIN. MICROBIOL.

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Page 5: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

CHARACTERIZATION OF COAGULASE-NEGATIVE STAPHYLOCOCCI 319

TABLE 3. Cellular fatty acid GLC correlation between seven different species of coagulase-negative staphylococcia

Mean correlation coefficient ± SD between speciesSpecies

S. epid (n = 178) S. war (n = 59) S. hom (n = 10) S. haem (n = 7) S. cap (n = 6) S. sim (n = 2) S. lugd (n = 2)

S. epid 87.27 ± 8.04 70.00 ± 11.27 76.60 ± 8.49 73.12 ± 11.1 77.43 ± 9.43 66.28 ± 5.74 71.46 ± 9.58S. war 91.24 ± 4.59 63.93 ± 9.79 86.76 ± 4.56 82.26 ± 5.32 70.16 ± 4.66 79.15 ± 4.27S. hom 85.54 ± 7.68 67.82 ± 7.47 75.05 ± 9.18 65.43 ± 6.82 67.61 ± 8.32S. haem 91.42 ± 3.82 82.54 ± 5.65 62.28 ± 5.87 81.30 ± 3.15S. cap 89.10 ± 5.27 68.62 ± 5.14 74.80 ± 2.68S. Sim 95.65 66.41 ± 3.26S. lugd 91.84

a Abbreviations: S. epid, S. epidermidis; S. war, S. warneri; S. hom, S. hominis; S. haem, S. haemolyticus; S. cap, S. capitis; S. sim, S. simulans; S. lugd,S. lugdunensis.

was compared to the clustering by the automated GLCanalysis, identical results were obtained in 244 of 264 cases(92.4%). Within the individual species, the agreement be-tween the methods was as follows: S. epidermidis, 168 of 178(94.4%); S. warneri, 55 of 59 (93.2%); S. hominis, 5 of 10(50.0%); S. haemolyticus, 6 of 7 (85.7%); S. capitis, 6 of 6(100.0%); S. simulans, 2 of 2 (100.0%); and S. lugdunensis,2 of 2 (100.0%). Identical results were obtained for all sevenreference strains.A number of incongruently positioned samples in the

various clusters was apparent. In the S. epidermidis cluster,7 of 175 (4%) fell into this category. In the S. warneri cluster,10 of 65 (15.4%) of the tested samples proved to be dispar-ately positioned. This particular cluster, as stated above,contained a subcluster of six S. haemolyticus samples.Alternatively, if this cluster was reorganized to include acombined S. warneri-S. haemolyticus cluster, the incidenceof variability in the positions of the samples would decreaseto 4 of 65 (6.15%). For the S. capitis (n = 6), the S. simulans(n = 2), and the S. hominis (n = 5) clusters, none of theorganisms were disparately positioned. As stated above,seven S. epidermidis and two S. warneri samples remainedoutside all clusters.

Correlation analysis of the multiple isolates from a givenpatient. Similarity indices between consecutive patient iso-lates were calculated for all 60 patients from whom multipleblood isolates were recovered. The purpose was to deter-mine whether the GLC profile correlation analysis couldestablish identity between those isolates which, in fact, wererepresentative of the same bacterial strain. Those multipleisolates from a given patient, determined to be identical bythe above-mentioned standard techniques, were operation-ally defined as the same strain. The correlation coefficients

S.epidermidisS.hominisS.warneriS.haemolyticusS.capitisS.lugdunensisS.simulans

0 50 100

FIG. 2. A dendrogram showing the GLC fatty acid profile simi-larities between seven species of coagulase-negative staphylococci.The dendrogram was established by cluster analysis of the mean

values of similarity indices. Interspecies similarity indices are indi-cated on the horizontal axis.

of the fatty acid profiles between the multiple patient isolatesare shown in Table 4. The correlation coefficients of themultiple identical blood isolates from 41 patients varied from92.79 to 99.01. In 90.2% (37 of 41) of the cases, thecorrelation value was >97; in 97.6% (40 of 41) of the cases,it was >95. The correlation values of the multiple, noniden-tical blood isolates from 19 patients varied from 59.97 to94.90. When the isolates were of the same species, thecorrelation value varied from 84.22 to 94.90 (mean, 89.99 +

3.87). The mean value of correlation between multiple noni-dentical S. epidermidis blood isolates (from 8 patients) was89.79 ± 4.04, which was quite comparable to the meancorrelation value of 87.27 + 8.04 (Table 3) between randomS. epidermis isolates. In contrast, a significantly (P < 0.001)higher mean correlation value between consecutive isolates,97.65 + 1.16, was detected in those 33 patients who hadmultiple identical S. epidermidis blood isolates.

DISCUSSION

Different species of coagulase-negative staphylococci canbe unequivocally distinguished when stringent DNA-DNAhybridization conditions are used (29). This methodology,however, is highly specialized and used primarily as aresearch tool. At present, no completely reliable systempractical as a routine diagnostic tool exists for identificationof coagulase-negative staphylococci to the species level.Commercially available panels, which identify on the basisof functional differences in the metabolic pathways, do notallow species identification with certainty (25). Thus, alter-native methods are being developed for speciation. Amongthe most interesting is whole-cell protein analysis; sodiumdodecyl sulfate-polyacrylamide gel electrophoresis has pro-vided species identification with a high degree of accuracy.By using this technique and immunoblotting, Thomson-Carter and Pennington (33) were able to characterize ninecoagulase-negative species. According to Pierre et al. (26),the penicillin-binding protein profiles provide an evengreater degree of species identification accuracy than doesexamination of the total solubilized proteins. Another inter-esting approach to species identification, the GLC analysisof cellular fatty acids, has indicated distinct quantitativedifferences between various coagulase-negative species, de-spite qualitative similarities (7, 21, 30).The purpose of this study was to further evaluate the

applicability of the GLC fatty acid analysis for speciesidentification of coagulase-negative staphylococci. Consis-tent with the previous findings (7, 8, 21, 30), no obviousqualitative differences in the fatty acid compositions wereobserved between the different coagulase-negative Staphy-

VOL. 29, 1991

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Page 6: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

320 KOTILAINEN ET AL.

-S. haemolyticus

-S. warneri

CS .capitis-S.lugdunensis

-S.epidermidis

simulanshominis

0 50 100FIG. 3. A dendrogram showing the GLC fatty acid profile simi-

larities between 264 isolates of coagulase-negative staphylococci.The isolates were identified by the API Staph-Trac and ATB 32STAPH procedures as seven different species: S. epidermidis, S.warneri, S. hominis, S. haemolyticus, S. capitis, S. simulans, and S.Iugdunensis. A weighted-pair cluster analysis grouped the isolatesinto species-specific clusters; the compositions of the seven clustersshown in the figure are explained in detail in the text. The fatty acid

profile correlation between individual isolates is indicated on thehorizontal axis.

lococcus species. Admittedly, a few fatty acids which pro-duced very low mean peak areas were not detected in allcoagulase-negative species studied. Additionally, differ-ences in the apparent fatty acid prevalences were observedbetween individual isolates of those species in which therelative mean amount of that particular fatty acid was lessthan 1%. It should be noted here that in such cases, theobserved differences most probably reflected a function ofthe detection limit of the method rather than any truedissimilarity. Although qualitative differences were not ob-served, various species differed significantly from each otherin the relative amounts of several individual fatty acids.Some of the species-specific characteristics were quite dis-tinctive. For example, S. epidermidis was characterized by arelatively large amount of the fatty acid C18:0, S. warneri wascharacterized by a large amount of the fatty acid Ca-15:0, 5haemolyticus was characterized by a large amount of thefatty acid Cai17:0, and S. capitis was characterized by a largeamount of the fatty acid Cji17:0. Still, there was too muchoverlapping in the quantitative differences between variousspecies to permit reliable species identification on the basisof the quantity of any single fatty acid.The computerized fatty acid pattern correlation analysis

did, however, separate the staphylococcal isolates into spe-cies-specific clusters, thereby allowing species identifica-tion. Two main clusters were formed by S. epidermidis andS. warneri, the species constituting the bulk of our straincollection. Although the number of isolates belonging to thefive other species was few, even these isolates were dividedinto species-specific clusters. Species identification by GLCclustering was quite consistent with species identification bybiochemical analysis; the results were similar in 92.4% of thecases. Moreover, correlation analysis demonstrated that theintraspecies similarities were higher than were the interspe-cies similarities. On the basis of the correlation values, themost distinct species were S. epidermidis, S. hominis, and S.simulans; the interspecies similarity indices between theseand all other species were <80. Collectively, these findingsindicate clear differences in the GLC fatty acid profilesbetween the seven coagulase-negative species studied. In-deed, our data, confirming earlier findings of others (7, 21,30), suggest that cellular fatty acid profile analysis may nowconstitute an alternative methodology for identification ofcoagulase-negative staphylococci to the species level.

Coagulase-negative staphylococci are the leading cause ofhospital-acquired bloodstream infections (27). Accordingly,in patients susceptible to coagulase-negative staphylococcalsepticemias, even a single positive blood isolation warrantscareful evaluation. Even so, coagulase-negative staphylo-cocci, as part of the normal skin flora, also constitute one ofthe most common contaminants of bacterial cultures. There-fore, a requirement for defining a specific strain as clinicallysignificant is its multiple recovery from two or more culturestaken from the patient at separate times.Of the various methods available for distinguishing indi-

vidual staphylococcal strains, no single method has provedtotally satisfactory. Thus, a combination of some of thesetechniques is often required for confirming the relatedness ofseparate isolates (3, 10, 14, 17, 23, 28). Because of theirready availability, antimicrobial susceptibility patterns are infrequent clinical use for strain characterization. The resis-tance of most nosocomial strains to multiple antibioticsappears to limit the usefulness of antibiograms (22, 25).Another limitation may be the unstable antibiotic resistanceproperty (25). Biotyping is reproducible; because of a mod-est number of different biotypes, however, it is often not

J. CLIN. MICROBIOL.

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Page 7: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

CHARACTERIZATION OF COAGULASE-NEGATIVE STAPHYLOCOCCI 321

TABLE 4. Correlation values of the fatty acid profiles betweenmultiple, consecutive blood isolates from 60 patients

Group and species Patient No. of Correlation

no. isolates value ± SD

Patients with multiple, identicalisolatesa

S. epidermidis

S. warneri

S. hominis

S. lugdunensis

S. capitis

Patients with multiple, noniden-tical isolates' belonging tothe same species

S. epidermidis

S. warneriS. hominis

1 2 97.972 2 97.373 3 97.15 ± 1.144 2 97.355 2 98.546 3 98.35 ± 0.367 3 97.79 ± 0.598 2 98.899 4 97.33 ± 1.6610 2 99.0111 3 98.49 ± 0.0912 3 98.20 ± 0.7813 3 95.11 ± 2.1914 4 97.63 ± 0.5315 2 98.3016 2 98.5017 2 97.4418 4 97.75 ± 0.9719 2 97.9620 2 97.7721 2 97.0922 2 97.9623 2 98.0124 2 98.0925 2 97.9226 2 95.8827 2 97.4828 2 98.0929 2 97.0030 2 98.7331 2 98.0232 2 98.5333 2 92.7934 3 98.20 ± 0.6635 2 98.2036 3 98.71 ± 0.2237 2 98.2938 2 96.2739 2 97.9140 2 97.7241 2 98.64

42 2 84.8643 2 86.8044 2 94.9045 2 93.8146 2 84.2247 2 94.6248 2 89.1349 2 90.0050 2 93.7751 2 87.80

Continued

sufficiently discriminatory (25). Phage typing lacks standard-ization (25). Serotyping is poorly developed because of thedifficulties of preparing specific antisera (22, 25). Moleculartechniques provide more useful typing methods, yet most ofthem remain too expensive and technically demanding forroutine diagnostics. Consequently, there remains a need to

TABLE 4-Continued

Patient No. of Correlationno. isolates value + SD

Patients with multiple isolates 52 2 71.82belonging to different 53 3 77.43 t 6.88speciesb 54 2 75.24

55 2 60.4456 2 76.0957 2 68.2558 2 60.2659 2 71.9260 2 59.97

a Identity was determined by antimicrobial susceptibility pattern, biotyp-ing, plasmid profile analysis, and the capacity to adhere and produce slime.

b From patients 52 to 60 were recovered the following: 52, S. epidermidisand S. hominis; 53, S. epidermidis, S. hominis, and S. warneri; 54, S.epidermidis and S. warneri; 55, S. epidermidis and S. warneri; 56, S.epidermidis and S. capitis; 57, S. epidermidis and S. hominis; 58, 5. warneriand S. hominis; 59, S. epidermidis and S. hominis; and 60, S. epidermidis andS. warneri.

develop supplementary methods for the epidemiologicaltyping of coagulase-negative staphylococci.The results reported in this study strongly suggest that

identity as well as differences between multiple staphylococ-cal blood isolates can be determined by the degree ofresemblance of their fatty acid profiles. For those multipleisolates determined to be identical by standard techniques(such as the antibiogram, biotype, and plasmid profile), thefatty acid profiles were similar. In contrast, dissimilar pro-files were observed for those isolates determined to benonidentical by the standard techniques. The correlation ofthe GLC fatty acid profiles was >95 for all except onepatient with multiple identical isolates. The correlation co-efficient was, on the other hand, <95 in all those patientswith multiple nonidentical isolates. The mean values of thecorrelations between these multiple nonidentical patientisolates, belonging to either the same or different species,were comparable with those between any random isolates,of either the same or different species.Compared with the standard typing methods referred to

above, the GLC profile correlation analysis has many appar-ent advantages. The procedure is cheap, quick, easy toperform, and reproducible (8). As such, the technique issuitable even for the rapid screening of a great number ofbacterial isolates. In this regard, the GLC assay would mostlikely prove to be useful for interrelating separate isolatesduring nosocomial outbreaks of coagulase-negative staph-ylococci. Furthermore, the GLC method may be especiallyvaluable for the typing of not only coagulase-negative staph-ylococcal isolates which lack plasmids but also those isolateswhich have lost some of their resistance plasmids and havecoincidentally changed their antimicrobial susceptibility pro-files. Moreover, the fatty acid profile correlation analysisoffers a practical method for establishing identity betweenthose clinically important strains which are untypable withphages.

In conclusion, the results presented herein indicate thatcoagulase-negative staphylococci can be identified to thespecies level by their GLC fatty acid profiles. Moreover,computerized fatty acid profile correlation analysis can beapplied for determining identity, as well as differences,between multiple coagulase-negative staphylococcal bloodisolates. Yet another application for this technique, foresee-able in the future, might be discrimination between epidemi-

VOL. 29, 1991

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

Page 8: Application ofGas-Liquid Chromatographic Analysis Fatty ... · JOURNALOFCLINICAL MICROBIOLOGY, Feb. 1991, p. 315-322 Vol. 29, No. 2 0095-1137/91/020315-08$02.00/0 Copyright © 1991,

322 KOTILAINEN ET AL.

ologically related and unrelated strains during hospital out-breaks caused by coagulase-negative staphylococci.

ACKNOWLEDGMENTS

We thank Kristi Tuomela, Marja-Riitta Terasjarvi, Tarja Boman,and Tarja Laine for expert technical assistance and Jeri L. Hill forcritical reading and revision of the language of the manuscript.

This work was supported by grants from the Academy of Finlandand Sigrid Juselius Foundation.

REFERENCES1. Almeida, R. J., and J. H. Jorgensen. 1983. Identification of

coagulase-negative staphylococci with API STAPH-IDENTsystem. J. Clin. Microbiol. 18:254-257.

2. Archer, G. L. 1984. Staphylococcus epidermidis: the organism,its diseases, and treatment, p. 25-48. In J. S. Remington andM. N. Swartz (ed.), Current clinical topics in infectious diseases5. McGraw-Hill Book Co., New York.

3. Christensen, G. D., J. T. Parisi, A. L. Bisno, W. A. Simpson, andE. H. Beachey. 1983. Characterization of clinically significantstrains of coagulase-negative staphylococci. J. Clin. Microbiol.18:258-269.

4. Christensen, G. D., W. A. Simpson, A. L. Bisno, and E. H.Beachey. 1982. Adherence of slime-producing strains of Staph-ylococcus epidermidis to smooth surfaces. Infect. Immun. 37:318-326.

5. Crouch, S. F., T. A. Pearson, and D. M. Parham. 1987.Comparison of modified Minitec system with Staph-Ident sys-tem for species identification of coagulase-negative staphylo-cocci. J. Clin. Microbiol. 25:1626-1628.

6. Doern, G. V., J. E. Earls, P. A. Jeznaih, and D. S. Parker. 1983.Species identification and biotyping of staphylococci by the APIStaph-Ident system. J. Clin. Microbiol. 17:260-263.

7. Durham, D. R., and W. E. Kloos. 1978. Comparative study ofthe total cellular fatty acids of Staphylococcus species of humanorigin. Int. J. Syst. Bacteriol. 28:223-228.

8. Eerola, E., and 0. Lehtonen. 1988. Optimal data processingprocedure for automatic bacterial identification by gas-liquidchromatography of cellular fatty acids. J. Clin. Microbiol.26:1745-1753.

9. Etienne, J., Y. Brun, N. El Solh, V. Delorme, C. Mouren, M.Bes, and J. Fleurette. 1988. Characterization of clinically signif-icant isolates of Staphylococcus epidermidis from patients withendocarditis. J. Clin. Microbiol. 26:613-617.

10. Etienne, J., F. Renaud, M. Bes, Y. Bryn, T. B. Greenland, J.Freney, and J. Fleurette. 1990. Instability of characteristicsamongst coagulase-negative staphylococci causing endocardi-tis. J. Med. Microbiol. 32:115-122.

11. Gemmel, C. G., and J. E. Dawson. 1982. Identification ofcoagulase-negative staphylococci with the API Staph system. J.Clin. Microbiol. 16:874-877.

12. Giger, O., C. C. Charilaou, and K. R. Cundy. 1984. Comparisonof the API Staph-Ident and DMS Staph-Trac systems withconventional methods used for identification of coagulase-neg-ative staphylococci. J. Clin. Microbiol. 19:68-72.

13. Hamory, B. H., and J. T. Parisi. 1987. Staphylococcus epider-midis: a significant nosocomial pathogen. Am. J. Infect. Control15:59-74.

14. Hartstein, A. I., M. A. Valvano, V. H. Morthland, P. C. Fuchs,S. A. Potter, and J. H. Crosa. 1987. Antimicrobic susceptibilityand plasmid profile analysis as identity tests for multiple bloodisolates of coagulase-negative staphylococci. J. Clin. Microbiol.25:589-593.

15. Karchmer, A. W., G. L. Archer, and W. E. Dismukes. 1983.Staphylococcus epidermidis causing prosthetic valve endocardi-tis: microbiologic and clinical observations as guides to therapy.

Ann. Intern. Med. 98:447-455.16. Lowy, F. D., and S. M. Hammer. 1983. Staphylococcus epider-

midis infections. Ann. Intern. Med. 99:834-839.17. Ludlam, H. A., W. C. Noble, R. R. Marples, and I. Phillips.

1989. The evaluation of a typing scheme for coagulase-negativestaphylococci suitable for epidemiological studies. J. Med.Microbiol. 30:161-165.

18. Maki, D. G. 1982. Infections associated with intravascular lines,p. 309-363. In J. S. Remington and M. N. Swartz (ed.), Currentclinical topics in infectious diseases 3. McGraw-Hill Book Co.,New York.

19. Moss, C. W., and 0. L. Nunez-Montiel. 1982. Analysis ofshort-chain acids from bacteria by gas-liquid chromatographywith a fused-silica capillary column. J. Clin. Microbiol. 15:308-311.

20. Nafziger, D. A., and R. P. Wenzel. 1989. Coagulase-negativestaphylococci. Epidemiology, evaluation, and therapy. Infect.Dis. Clin. North Am. 3:915-929.

21. O'Donnel, A. G., M. R. Nahaie, M. Goodfellow, D. E. Minnikin,and V. Huijek. 1985. Numerical analysis of fatty acid profiles inthe identification of staphylococci. J. Gen. Microbiol. 131:2023-2033.

22. Parisi, J. T. 1985. Coagulase-negative staphylococci and theepidemiological typing of Staphylococcus epidermidis. Micro-biol. Rev. 49:126-139.

23. Parisi, J. T., B. C. Lampson, D. L. Hoover, and J. A. Khan.1986. Comparison of epidemiologic markers for Staphylococcusepidermidis. J. Clin. Microbiol. 24:56-60.

24. Patrick, C. C. 1990. Coagulase-negative staphylococci: patho-gens with increasing clinical significance. J. Pediatr. 116:497-507.

25. Pfaller, M. A., and L. A. Herwaldt. 1988. Laboratory, clinical,and epidemiological aspects of coagulase-negative staphylo-cocci. Clin. Microbiol. Rev. 1:281-299.

26. Pierre, J., L. Gutman, M. Bornet, E. Bergogne-Berezin, and R.Williamson. 1990. Identification of coagulase-negative staphylo-cocci by electrophoretic profile of total proteins and analysis ofpenicillin-binding proteins. J. Clin. Microbiol. 28:443-446.

27. Ponce De Leon, S., and R. P. Wenzel. 1984. Hospital-acquiredbloodstream infections with Staphylococcus epidermidis. Am.J. Med. 77:639-644.

28. Renaud, F., J. Freney, J. Etienne, M. Bes, Y. Brun, 0. Barsotti,S. Andre, and J. Fleurette. 1988. Restriction endonucleaseanalysis of Staphylococcus epidermidis DNA may be a usefulepidemiological marker. J. Clin. Microbiol. 26:1729-1734.

29. Schleifer, K. H. 1986. Taxonomy of coagulase-negative staph-ylococci, p. 11-26. In P. A. Mardh and K. H. Schleifer (ed.),Coagulase-negative staphylococci. Almqvist & Wiksell Interna-tional, Stockholm.

30. Schleifer, K. H., and R. M. Kroppenstedt. 1990. Chemical andmolecular classification of staphylococci. J. Appl. Bacteriol.Symp. Suppl. 69:9S-24S.

31. Sneath, P. H. A., and R. R. Sokal. 1973. The estimation oftaxonomic resemblance, p. 114-187. In D. Kennedy and R. D.Park (ed.), Numerical taxonomy. W. H. Freeman & Co., SanFrancisco.

32. Sneath, P. H. A., and R. R. Sokal. 1973. Taxonomic structure, p.288-308. In D. Kennedy and R. D. Park (ed.), Numericaltaxonomy. W. H. Freeman & Co., San Francisco.

33. Thomson-Carter, F. M., and T. H. Pennington. 1989. Charac-terization of coagulase-negative staphylococci by sodium dode-cyl sulfate-polyacrylamide gel electrophoresis. J. Clin. Micro-biol. 27:2199-2203.

34. Wade, J. C., S. C. Schimpif, K. A. Newman, and P. H. Wiernik.1982. Staphylococcus epidermidis: an increasing cause of infec-tion in patients with granulocytopenia. Ann. Intern. Med.97:503-508.

J. CLIN. MICROBIOL.

on Decem

ber 15, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from