- Email: [email protected]
A comparison of methods to determine whether clinical isolates of Staphylococcus epidermidis from the same patient are related
Journal
of Hospital
Infection
(1996)
34, 31-42
A comparison of methods to determine whether clinical isolates of Staphylococcus epidermidis from the same patient are related G. Hedin
Department
of Clinical
Received 29 November
Microbiology,
University Sweden
Hospital,
1995; revised manuscript
S-751 85 Uppsala,
accepted 7 February
1996
Summary:
Staphylococcus epidennidis is a major cause of hospital-acquired infections but also part of the normal skin flora. A common clinical question is whether repeated isolation of S. epidermidis from one patient represents the same strain; because if different strains are isolated, they are often thought to be contaminants. In this study, different typing methods were compared to answer this question. Twenty isolates of S. epidermidis from five different patients were investigated. The isolates from each patient had identical or very similar antibiograms, and were recovered on different occasions. Typing was performed by antibiogram, biotype, slime production, plasmid profile, and pulsed-field gel electrophoresis (PFGE) banding pattern of SmaI digests of chromosomal DNA. In addition, the level of resistance to methicillin was determined by growth curves in broth containing methicillin for a series of different inocula for each isolate. The results showed that the isolates from each patient belonged to the same clone, but examples of instabilities in their antibiograms, plasmid profiles, as well as their PFGE banding patterns were seen. A change in the level of methicillin resistance was observed in one strain; otherwise this characteristic was found to be strain-specific and stable,in vivo. It was concluded that in combination with biotyping and antibiotic resistance testing the level of resistance to methicillin could be used as an aid to distinguish between two or more clinical isolates of S. epidermidis from the same patient. Keywords:
Staphylococcus
epiderrnidis;
bacteraemia;
typing;
identical;
growth
curve.
Introduction
Staphylococcus
epidermidis is part of the normal skin flora but is also an important pathogen among patients in hospital.’ In clinical practice it is often difficult to establish whether a S. epidermidis strain isolated from a patient is of clinical significance. If the same strain is repeatedly isolated from an infected patient, this suggests that it causes the infection, whereas if different strains are isolated, they are often thought to be contaminants.2 Correspondence 83 &tersund, 01956701/96/090031+12
to: Dr Sweden
G. Hedin
at Department
of Clinical
$12.00/O
Microbiology,
&tersund
Hospital,
0 1996 The Hospital
31
Infection
S-831
Soctety
G. Hedin
32
Different isolates with identical antibiograms indicate that it is probably the same strain. However, if in addition more discriminating typing methods are used, the isolates may ultimately prove to be unrelated.3’4 The aim of this study was to evaluate the use of different typing methods to determine whether two or more isolates from the same patient were identical. At the University Hospital, Uppsala, a prospective search for clinical isolates of S. epidermidis with identical antibiograms from the same patient was performed. Five patients were found who suffered from serious infections caused by S. epidermidis, i.e., isolates with the same antibiogram were repeatedly isolated and no other pathogens were found. Twenty isolates from these five patients were used in this study. Six methods were used for typing: (1) antibiogram, (2) biotyping, (3) a slime production test, (4) plasmid profile analysis, (5) pulsed-field gel electrophoresis (PFGE) of chromosomal DNA, and (6) growth curves in broth containing methicillin. The latter method for typing of S. epidermidis was evaluated for the first time in this study. The rationale of using this method is that resistance to methicillin in S. epidermidis is often expressed only by a minority of the cells in a bacterial population.‘-’ Tomasz et al., who worked mainly with Staphylococcus aureus, but also with S. epidermidis, described how the level of methicillin resistance varied from one strain to another but was specific for each strain and was a stable characteristic in vitro.’ Each strain consists of several subpopulations of cells that differ in their level of resistance to methicillin.’ Thus, if the level of resistance is also stable in z)izIo, it may be useful as a strain epidemiological marker. Materials
Bacterial
and methods
isolates
Twenty clinical isolates of S. epidermidis were used; their antibiograms are given in Table 1. They were isolated from five patients (A-E) and kept at -70°C until use. The clinical details of these patients are described below:
Patient
A. Two isolates, Al and A2, were recovered from a 64-year-old man who underwent surgery for a dissecting aneurysm of the ascending aorta. An aortic valve prosthesis was implanted. Because of postoperative fever, cefuroxime was administered, though without effect. Al was isolated from an inguinal wound 11 days after the operation, and A2 from a blood culture 14 days postoperation. Ultimately, vancomycin treatment cured the infection.
Patient
B. Isolates Bl and B2 were cultured from blood and tip of central venous catheter of a 63-year-old man with aortic stenosis and ischaemic heart disease. An aortic valve prosthesis was implanted. Several complications occurred postoperatively, including prolonged fever. Isolate Bl was recovered 11 days after the operation, and B2 when the catheter was removed 15 days postoperation.
S. epidermidis
typing
methods
33
Patient C. Seven isolates, Cl-C7, were isolated from the cerebrospinal fluid (CSF) of a two-month-old girl who was born with Arnold-Chiari’s malformation. A ventriculo-peritoneal shunt had been inserted when she was one month old because of hydrocephalus. After the operation she was febrile. S. epidermidis, Cl-C7, were repeatedly isolated from cultures of the CSF taken 17-31 days after the implantation of the shunt, in spite of treatment with co-trimoxazole, gentamicin and dicloxacillin. Patient D. Isolates Dl-D4 were recovered from a 40-year-old man who was operated on because of an acute subarachnoid haemorrhage. A ventricular catheter was inserted for drainage. One month later this was replaced by a ventriculo-peritoneal shunt. Dl was isolated from a culture of the CSF taken when the shunt was inserted. Nineteen days later an isolate with an identical antibiogram, D3, was isolated from the CSF, together with another isolate, D4, with a different antibiogram. Before that, 12 days after the first positive culture, isolate D2, which had an antibiogram identical to that of D4, was recovered from the CSF. As a result of the infection the shunt did not function well and had to be removed. Patient E. Five isolates, El-ES, were isolated from a 48-year-old man with acute myeloid leukaemia. One week after the patient had been given cytostatic drugs for the first time, he developed a high fever, and S. epidermidis was isolated from three blood cultures. El was the first of these, and E2 the last, was taken eight days later. Vancomycin was given for 21 days, first alone, then together with amikacin and co-trimoxazole; finally, ciprofloxacin was administered perorally. The patient recovered from the infection and a complete remission was seen in the bone marrow. However, one month later, after treatment with cytostatic drugs for the second time, the patient again became febrile. An S. epidermidis strain with an identical antibiogram as the earlier two isolates, except for resistance to fusidic acid, was isolated from a blood culture. Vancomycin was given for 10 days, with improvement in clinical status, but the central venous catheter was not removed. After two more cytostatic drug treatments, the patient was prepared for an autologous bone marrow transplantation which he underwent six months after the first treatment with cytostatic drugs. Before the transplantation, S. epidermidis was again isolated from three different blood cultures through the central venous catheter, although the patient had no fever. Isolate E4 was the first of these and was recovered nine days before ES. The antibiograms of these isolates were identical to that of E3. The catheter was removed and a new one inserted while vancomycin was being given. Typing methods Antibiogram. Standardized disc diffusion susceptibility formed on Isosensitest agar (Unipath Ltd, Basingstoke, biotics tested are shown in Table I.
tests’ were perUK). The anti-
34
G. Hedin
Biotyping. Biotyping was performed with a commercial kit for species t.yping of staphylococci, ‘API Staph’ (API Bio Merieux SA. Lyon, France).
Slimeproduction.
Slime production was determined by the tube adherence test described by Christensen et ~1.” Slime production was recorded as negative (-), weak (+), moderate (2+) or strong (3+) according to the density of the adherent biofilm.
Plasmid
profile. Plasmids lysis method.” They were and stained with ethidium and photographed. Phage molecular size marker. PFGE
were isolated by a modification separated on a 0.7% agarose gel bromide before being visualized h DNA digested with Hind111
of the alkaline at 80 V for 1 h with UV light was used as a
of chromosomal DNA. Bacterial colonies from blood agar were suspended in 0.5 ml PIV buffer (1 M NaCl, 10 mM Tris-HCl, pH 8.0) containing 250 ltg lysostaphin (Sigma, St. Louis, USA) and 2 mg lysozyme (Sigma). An equal volume of TE buffer ( 10 mM Tris-HCl, 0.1 mM EDTA, pH 7.5) containing 2% agarose (Sigma Type VII: low gelling temperature) at 55°C was then added. The mixture was immediately poured into the slots of a plastic mould (2 x 5 x 10 mm), allowed to solidify at room temperature and incubated at 37°C for 2 h. The agarose plugs were transferred to tubes each containing 3 mL lysis buffer (6 mM Tris-HCl, 10 mM EDTA, 0.2% deoxycholate, 0.5 lauroyl sarcosine, pH 8.0) with RNase (50 l.rg/ mL)(Boechringer GmbH, Mannheim, Germany) and incubated at 37°C for 2 h. The lysis buffer was replaced by proteolysis buffer (0.5 M EDTA, 1% lauroyl sarcosine) containing 1 mg/mL proteinase K (Boehringer). After overnight incubation at 50°C the agarose plugs were washed with distilled water and then with 2 mL TE buffer to which 30 PL of a 100 mM solution of phenylmethylsuphonyl fluoride in isopropanol was added. After incubation at room temperature for 1 h, three washing steps followed, first with distilled water, then two washes with TE buffer. The agarose plus were stored in TE buffer at 4°C until the next day. The restriction enzyme SmaI (Gibco BRL, Gaithersburg, USA) was used, each agar plug being incubated at room temperature for 4 h in a tube containing 100 U of the enzyme in 100 FL buffer. Small fragments of each agarose plug were placed into the slots of a 1% agarose gel and the slots sealed with 1% molten agarose (Sigma type II. Medium EEO). The DNA restriction fragments were separated by PFGE in a CHEF-DR II system (Bio-Rad Laboratories, Nazareth, Belgium). Electrophoresis was carried out at 180 V in 0.5 x TBE buffer (0.045 M Tris-borate, 0.001 M EDTA) with alternating pulses at a 120” angle in a 7-12 s pulse time gradient for 11 h and then a 20-40 s gradient for 13 h at 14°C. Pharmacia h DNA-PFGE marker (Pharmacia, Uppsala, Sweden) was used as a molecular size marker. The gels were stained with ethidium bromide and photographed under UV light.
S. epidermidis Table
I. Antibiograms
of 20 isolates
typing
of Staphylococcus
Antibiotic
Benzylpenicillin Methicillin Co-trimoxazole Doxycycline Erythromycin Clindamycin Chloramphenicol Rifampicin Vancomycin Ciprofloxacin Fusidic acid Gentamicin Netilmicin Tobramycin Amikacin
methods
35
epidermidis Isolate
from
Bl-B2
Cl-C7
Dl,D3
D2,D4
R R R
R R R S
R R R s
:: R S
:: R S R
E R
: S
: R
ii S R
: S
:
:
: S
: ::
: R R
: R R
:: R
patients
No.
Al-A2
ii S
jive
:: R R
El-E2
E3-E5
R
R R
:: R S
:: z
:: S z R R R R R
: S S R S
::
: R R
: R R
: S
Level of resistance to methicillin. The ‘Bioscreen’ analysing system (Labsystems, Helsinki, Finland) was used for continuous measurement of bacterial growth in broth containing methicillin.12 Bacterial colonies from the overnight growth on blood agar was suspended in 2 mL of phosphatebuffered saline and the cell density was adjusted to 10’ cfu/mL. This suspension was serially diluted lo-fold in Mueller Hinton broth (Difco Laboratories, Detroit, USA) supplemented with 2% NaCl, calcium, 50 mg/ L, magnesium, 25 mg/L, and methicillin, 16 mg/L. The dilutions (0.3 mL volumes) were transferred to the wells of specially designed microplates supplied by the manufacturer (Labsystems). For each strain, five different inocula were tested: 103, 104, 1OS,1O6and 1O7cfu, each generating a different growth curve in the ‘Bioscreen’ system. Growth was observed first in the wells with the highest inoculum sizes and later in the wells with the lower inoculum sizes. Strains expressing heterogeneous resistance to methicillin showed growth within 24 h only in the wells with the third or fourth highest inoculum sizes. The growth curves of different isolates were judged by the naked eye, and those isolates with similar curves were grouped together. This was repeated at least three times for each strain. Results Three of five patients (A, B and C) each yielded isolates of S. epidermidis with identical antibiograms. Of the remaining two patients, patient D harboured isolates with two different antibiograms and so did patient E. Table I shows that in total seven antibiogram patterns were identified and variability in resistance was most often seen with chloramphenicol and
36 Table
G. Hedin II.
Patient isolate Al ,A2 Bl.B2
a-c7 Dl D3 D2,D4 El E2 E3 E4,45
Summary
of the results
Antibiogram profile
of different
Slime production
A
1+
:
2+ 3+
E
2
E F F
i::
E
1+
methods Plasmid pattern
for
strain
characterization
PFGE profile I I II III III IV I I V VI
2+ 2+
Methicillin sensitivity* 1 : ;
4 4 2 ;
*See text.
fusidic acid. All isolates were uniformly resistant to amikacin and tobramycin but resistance to gentamicin and netilmicin was variable. Table II summarizes the results of other characteristics tested for the isolates. All isolates were uniform in biotype and all produced slime but to varying degrees. Slime production was strongest for all isolates from patient C and two from patient D. Seven plasmid patterns and six DNA macrorestriction profiles were found for all isolates. Only four levels of methicillin resistance were identified. Within each patient, the combination of methods showed that those from patients A, B and C were uniform indicating infection by a single strain unique for each patient. However, three strain phenotypes were found among isolates from patient D. Isolates Dl and D3 were identical in all respects except for the presence in Dl of a plasmid of ~2000 bps, which was present also in D2 and D4. The latter were indistinguishable from each other but differed from other isolates although some shared characters such as plasmid content, response to slime production test and level of resistance to methicillin were observed. The isolates from patient E showed the greatest heterogeneity in typing markers. These were characterized by two plasmid patterns, three PFGE profiles, and two responses to methicillin. Resistance to methicillin was expressed by all cells in the population in isolate El, whereas in the remaining isolates from this patient resistance was expressed by only a minority of cells. Discussion
In this study different typing methods were used to answer a common clinical question of whether isolates of S. epidermidis from one and the same patient are identical. Twenty S. epidermidis isolates from five different patients were selected on the basis that they had identical or closely similar antibiograms.
S. epidermidis
typing
methods
37
(b)
(a)
Figure 1. Plasmid patterns of Staphylococcus epidermidis strains isolated from patients A, B, D and E. (a) Lane 1: molecular size marker (Hind111 digest of h DNA); lane 2: Al; lane 3: A2; lane 4: Bl; lane 5: B2; lane 6: Dl; lane 7: D2; lane 8: D3; lane 9: D4. (b) Lane 1: molecular size marker; lane 2: El; lane 3: E2; lane 4: E3; lane 5: E4; lane 6: ES.
1
2
3
4
5
6
Figure 2. Plasmid patterns of Staphylococcus epidermidis Lane 1: molecular size marker (Hind111 digest of h DNA); C3; lane 5: C4; lane 6: C5; lane 7: C6; lane 8: C7.
7
strains lane
8
isolated from patient C. 2: Cl; lane 3: C2; lane 4:
It is known that the characteristics of a strain, even those commonly used for epidemiologic typing, may become modified over time. A given strain of S. epidermidis may acquire or lose plasmids and/or antibiotic resistance genes as observed in studies of strains isolated from the same patient on different occasions.13-15 PFGE patterns of these strains may undergo also slight modifications with time.16 Despite these reservations, the antibiogram and plasmid content of some strains of S. epidermidis may remain stable for several months.17 The isolates of S. epidermidis used in this study were recovered from the patients at time intervals of three, four,
G. Hedin 12345678
9 10 11 12 13 14
Figure 3. Pulsed-field gel electrophoresis of SmaI digests of chromosomal DNA. All different banding patterns for 20 isolates of Staphylococcus epidermidis are shown. As the patterns of all the C strains were identical, only that of C7 is shown and as the pattern of Al was identical to that of A2, Bl, B2, El and E2, only those of Al, El and E2 are shown. Lane 1: C7; Lane 2: molecular size marker (Pharmacia h DNA-PFGE marker: X1857 Sam 7 ligated concatemers, sizes 48.5, 97.0, 145.5 kb . . . etc.); lane 3: Dl; lane 4: D2; lane 5: D3; lane 6: D4; lane 7: molecular size marker; lane 8: Al; lane 9: El; lane 10: E2; lane 11: E3; lane 12: E4; lane 13: ES; lane 14: molecular size marker.
24 and 19 days for patients A-D, and with a time interval of six months in patient E. Overall the 20 isolates exhibited a single biotype, seven antibiograms, seven plasmid profiles, six PFGE patterns and four levels of resistance to methicillin. Isolates identical according to one typing method were not invariably grouped together by another method and so the combination of results identified 10 groups of strains. Biotyping has a low discriminatory power. The majority, 70% or more of blood culture isolates of coagulase-negative staphylococci, belong to the species S. epidermidis, most of which have the same biotype.” Nevertheless biotyping should be included in a typing scheme to make identification of an unusual biotype possible. Both homogeneous and heterogeneous expression of methicillin resistance was displayed by the isolates. Those that expressed heterogeneous resistance produced different growth curves. Four types of growth curve response to methicillin were observed in the present study, which may be compared with the four arbitrary expression classes proposed by Tomasz et aZ.* for strains of staphylococci. The two isolates from patient A were identical by all the methods used, as were those from patient B and patient C. The A and B isolates came
S. e$idermidis
typing
0
39
methods
12
24
0.8
I 0
12
24 Hours
Figure 4. Growth curves in broth containing methicillin for the 20 isolates of S. epidermidis. The growth curves for each of five inoculum sizes are shown in the same figure. The growth curves of strains isolated from the same patient considered as identical are superimposed on one another in the figures. (a) Al and A2. (b) Bl and B2. (c) D2 and D4. (d) Dl and D3. (e) El. (f) E2-ES. (g) Cl-C7.
from the same department and had identical PFGE patterns and levels of resistance to methicillin. The A isolates were, however, susceptible to fusidic acid, whereas the B isolates were resistant, the latter also contained an additional plasmid and produced slightly more slime. The PFGE pattern
40
G. Hedin
of the A and B strains was also identical to that of the first and second strains from patient E. The unreliability of plasmid profiling as a sole method for epidemiologic typing was illustrated by the results on isolates from patient D. The antibiograms, PFGE patterns, slime production and level of resistance to methicillin all suggested that Dl and D2 were identical to each other and different from D3 and D4, which in turn were identical to each other. The plasmid profiles of the strains, however, suggested that Dl, D2 and D4 were identical, while D3, where a small plasmid was missing, was different. The strains isolated from patient E were interesting also because six months had elapsed between the isolation of the first and last strain. Genetic changes among the strains were not therefore completely unexpected. The typing results suggested that a continuous genetic change had occurred within a number of strains with a common origin, i.e., a clone. The difference between the first two strains isolated from patient E was subtle. These strains had identical antibiograms, plasmid and PFGE patterns, and slime production. The only difference between them was that in El, methicillin resistance was expressed homogeneously, whereas in E2 it was expressed heterogeneously. This was the only case in the present study where the level of resistance to methicillin was found to be unstable in viva. The third strain, E3, was isolated one month later. This strain had acquired resistance to fusidic acid and the PFGE pattern also differed slightly from that of El and E2. Plasmid pattern and slime production, however, were identical to those of the earlier strains. According to the guidelines for strain typing by PFGE recently proposed by Tenover et ~1.‘~ the E3 strain was classified as possibly related to El and E2. The restriction pattern of El and E2 contains two bands (sizes approximately 390 and 110 kb), which are missing in the restriction pattern of E3, where instead two other bands (sizes approximately 240 and 260 kb) are present. In fact, a change in the position of only one restriction site may explain the observed differences. The last two strains, E4 and ES, were isolated five months later. They had the same antibiogram as the earlier strains, but an additional plasmid was seen, the slime production was slightly less profuse, and the PFGE pattern was also slightly modified. The level of resistance to methicillin was identical in strains E2-ES. In summary, examples were seen in this study where one typing method indicated a difference between two isolates, though a relationship was clearly indicated by the other methods, i.e., the strains belonged to the same clone. Accordingly, the plasmid profile was observed to change within a clone of strains from a patient, as was the PFGE pattern and the level of resistance to methicillin within a clone from a second patient. Identical PFGE patterns were found also in isolates that were not epidemiologically related. Thus no single method can provide the answer to the question of the relatedness of clinical isolates. Ultimately, the most reliable result will be obtained if different methods are used in combination. In the present
5’. epidermidis
typing
methods
41
study, identical antibiograms among isolates of S. epidermidis from the same patient indicated always that the strains were closely related, but this may not always be the case.4 It seems, therefore, that in clinical practice the use of antibiograms for epidemiologic typing is a natural first step. The level of resistance to methicillin is stable in viva in strains of S. epidermidis, as shown in this study, and therefore, can be used as an additional aid to identifying strains. The pulsed-field gel electrophoresis Copenhagen, Denmark, by the author. Seruminstitut, for her assistance.
studies were I am grateful
performed at Statens to Dr B. Thamdrup-Rosdahl,
Seruminstitut, Statens
References 1. Pfaller MA, Herwaldt LA. Laboratory, clinical, and epidemiological aspects of coagulase-negative staphylococci. Clin Micvobiol Rev 1988; 1: 281-299. 2. Archer GL. Coagulase-negative staphylococci in blood cultures: the clinician’s dilemma. Infect Control 1985; 6: 477478. 3. Ludlam HA, Noble WC, Marples RR, Philips I. The -evaluation of a typing scheme for coagulase-negative staphylococci suitable for epidemiological studies. J Med Microbiol 1989; 30: 161-165. 4. Khatib R, Reiderer KM, Clark JF, Khatib S, Briski LE, Wilson FM. Coagulasenegative staphylococci in multiple blood cultures: strain relatedness and determinants of same-strain bacteremia. J Clin Microbial 1995; 33: 816-820. 5. Chambers HF. Coagulase-negative staphylococci resistant to p-lactam antibiotics in vivo produce penicillin-binding protein 2a. Antimicrob Agents Chemother 1987; 31: 1919-1924. 6. Chambers HF. Methicillin-resistant staphylococci. Clin Microbial Rev 1988; 1: 173-186. BJ, Tomasz A. Expression of methicillin resistance in heterogeneous strains 7. Hartman of Staahvlococcus aureus. Antimicrob Agents Chemother 1986: 29: 85-92. 8. Tomasz *A, Nachman S, Leaf H. Stableclasses of phenotypid expression in methicillinAntimicrob Agents Chemother 1991; 35: resistant clinical isolates of staphylococci. 124-l 29. 9. Swedish Reference Group for Antibiotics. The Swedish SIR-system for susceptibility testing of bacteria. SGRA 1991. Stockholm. 10. Christensen GD, Andrew-Simpson W, Bisno AL, Beachey EH. Adherence of slimeproducing strains of Staphylococcus epidermidis to smooth surfaces. Infect Immun 1982; 37: 318-326. HC, Doly J. A rapid alkaline extraction procedure for screening recombinant 11. Birnboim plasmid DNA. Nucleic Acids Res 1979; 7: 1513-1523. G, Hambraeus A. Screening tests for the detection of methicillin resistance in 12. Hedin Staphylococcus epidermidis. J Antimicrob Chemother 1991; 28: 681-694. 13. Etienne J, Renaud F, Bes M et al. Instability of characteristics amongst coagulasenegative staphylococci causing endocarditis. J Med Microbial 1990; 32: 115-l 22. H, Jask D, Hammerberg 0. Evaluation of restriction endo14. Bialkowska-Habrzanska nuclease fingerprinting of chromosomal DNA and plasmid profile analysis for characterization of multiresistant coagulase-negative staphylococci in bacteremic neonates. J Clin Microbial 1990; 28: 269-275. PA, Plorde JJ, Gordon KP et al. Instability of antibiotic resistance in a strain 15. Mickelsen of Staphylococcus epidermidis isolated from an outbreak of prosthetic valve endocarditis. J Infect Dis 1985; 152: 50-58. 16. Goering RV, Winters MA. Rapid method for epidemiological evaluation of Grampositive cocci by field inversion gel electrophoresis. J Clin Microbial 1992; 30: 577-580. G, Hambraeus A. Enhanced ability to colonize the skin: a possible explanation 17. Hedin for the epidemic spread of certain strains of Staphylococcus epidermidis. J. Hosp Infect 1993; 25: 251-264.
42 18.
G. Hedin Hedin G. Comparison of coagulase-negative
of genotypic staphylococcal
and phenotypic isolates from
methods for blood cultures.
species
APMIS
identification
1994; 102:
855-864. 19.
Tenover patterns
FC, Arbeit, RD, Goering, produced by pulsed-field
J Clin Microbial.
RV et al. Interpreting gel electrophoresis:
1995; 33: 2233-2239.
chromosomal DNA restriction criteria for bacterial strain typing.
of Hospital
Infection
(1996)
34, 31-42
A comparison of methods to determine whether clinical isolates of Staphylococcus epidermidis from the same patient are related G. Hedin
Department
of Clinical
Received 29 November
Microbiology,
University Sweden
Hospital,
1995; revised manuscript
S-751 85 Uppsala,
accepted 7 February
1996
Summary:
Staphylococcus epidennidis is a major cause of hospital-acquired infections but also part of the normal skin flora. A common clinical question is whether repeated isolation of S. epidermidis from one patient represents the same strain; because if different strains are isolated, they are often thought to be contaminants. In this study, different typing methods were compared to answer this question. Twenty isolates of S. epidermidis from five different patients were investigated. The isolates from each patient had identical or very similar antibiograms, and were recovered on different occasions. Typing was performed by antibiogram, biotype, slime production, plasmid profile, and pulsed-field gel electrophoresis (PFGE) banding pattern of SmaI digests of chromosomal DNA. In addition, the level of resistance to methicillin was determined by growth curves in broth containing methicillin for a series of different inocula for each isolate. The results showed that the isolates from each patient belonged to the same clone, but examples of instabilities in their antibiograms, plasmid profiles, as well as their PFGE banding patterns were seen. A change in the level of methicillin resistance was observed in one strain; otherwise this characteristic was found to be strain-specific and stable,in vivo. It was concluded that in combination with biotyping and antibiotic resistance testing the level of resistance to methicillin could be used as an aid to distinguish between two or more clinical isolates of S. epidermidis from the same patient. Keywords:
Staphylococcus
epiderrnidis;
bacteraemia;
typing;
identical;
growth
curve.
Introduction
Staphylococcus
epidermidis is part of the normal skin flora but is also an important pathogen among patients in hospital.’ In clinical practice it is often difficult to establish whether a S. epidermidis strain isolated from a patient is of clinical significance. If the same strain is repeatedly isolated from an infected patient, this suggests that it causes the infection, whereas if different strains are isolated, they are often thought to be contaminants.2 Correspondence 83 &tersund, 01956701/96/090031+12
to: Dr Sweden
G. Hedin
at Department
of Clinical
$12.00/O
Microbiology,
&tersund
Hospital,
0 1996 The Hospital
31
Infection
S-831
Soctety
G. Hedin
32
Different isolates with identical antibiograms indicate that it is probably the same strain. However, if in addition more discriminating typing methods are used, the isolates may ultimately prove to be unrelated.3’4 The aim of this study was to evaluate the use of different typing methods to determine whether two or more isolates from the same patient were identical. At the University Hospital, Uppsala, a prospective search for clinical isolates of S. epidermidis with identical antibiograms from the same patient was performed. Five patients were found who suffered from serious infections caused by S. epidermidis, i.e., isolates with the same antibiogram were repeatedly isolated and no other pathogens were found. Twenty isolates from these five patients were used in this study. Six methods were used for typing: (1) antibiogram, (2) biotyping, (3) a slime production test, (4) plasmid profile analysis, (5) pulsed-field gel electrophoresis (PFGE) of chromosomal DNA, and (6) growth curves in broth containing methicillin. The latter method for typing of S. epidermidis was evaluated for the first time in this study. The rationale of using this method is that resistance to methicillin in S. epidermidis is often expressed only by a minority of the cells in a bacterial population.‘-’ Tomasz et al., who worked mainly with Staphylococcus aureus, but also with S. epidermidis, described how the level of methicillin resistance varied from one strain to another but was specific for each strain and was a stable characteristic in vitro.’ Each strain consists of several subpopulations of cells that differ in their level of resistance to methicillin.’ Thus, if the level of resistance is also stable in z)izIo, it may be useful as a strain epidemiological marker. Materials
Bacterial
and methods
isolates
Twenty clinical isolates of S. epidermidis were used; their antibiograms are given in Table 1. They were isolated from five patients (A-E) and kept at -70°C until use. The clinical details of these patients are described below:
Patient
A. Two isolates, Al and A2, were recovered from a 64-year-old man who underwent surgery for a dissecting aneurysm of the ascending aorta. An aortic valve prosthesis was implanted. Because of postoperative fever, cefuroxime was administered, though without effect. Al was isolated from an inguinal wound 11 days after the operation, and A2 from a blood culture 14 days postoperation. Ultimately, vancomycin treatment cured the infection.
Patient
B. Isolates Bl and B2 were cultured from blood and tip of central venous catheter of a 63-year-old man with aortic stenosis and ischaemic heart disease. An aortic valve prosthesis was implanted. Several complications occurred postoperatively, including prolonged fever. Isolate Bl was recovered 11 days after the operation, and B2 when the catheter was removed 15 days postoperation.
S. epidermidis
typing
methods
33
Patient C. Seven isolates, Cl-C7, were isolated from the cerebrospinal fluid (CSF) of a two-month-old girl who was born with Arnold-Chiari’s malformation. A ventriculo-peritoneal shunt had been inserted when she was one month old because of hydrocephalus. After the operation she was febrile. S. epidermidis, Cl-C7, were repeatedly isolated from cultures of the CSF taken 17-31 days after the implantation of the shunt, in spite of treatment with co-trimoxazole, gentamicin and dicloxacillin. Patient D. Isolates Dl-D4 were recovered from a 40-year-old man who was operated on because of an acute subarachnoid haemorrhage. A ventricular catheter was inserted for drainage. One month later this was replaced by a ventriculo-peritoneal shunt. Dl was isolated from a culture of the CSF taken when the shunt was inserted. Nineteen days later an isolate with an identical antibiogram, D3, was isolated from the CSF, together with another isolate, D4, with a different antibiogram. Before that, 12 days after the first positive culture, isolate D2, which had an antibiogram identical to that of D4, was recovered from the CSF. As a result of the infection the shunt did not function well and had to be removed. Patient E. Five isolates, El-ES, were isolated from a 48-year-old man with acute myeloid leukaemia. One week after the patient had been given cytostatic drugs for the first time, he developed a high fever, and S. epidermidis was isolated from three blood cultures. El was the first of these, and E2 the last, was taken eight days later. Vancomycin was given for 21 days, first alone, then together with amikacin and co-trimoxazole; finally, ciprofloxacin was administered perorally. The patient recovered from the infection and a complete remission was seen in the bone marrow. However, one month later, after treatment with cytostatic drugs for the second time, the patient again became febrile. An S. epidermidis strain with an identical antibiogram as the earlier two isolates, except for resistance to fusidic acid, was isolated from a blood culture. Vancomycin was given for 10 days, with improvement in clinical status, but the central venous catheter was not removed. After two more cytostatic drug treatments, the patient was prepared for an autologous bone marrow transplantation which he underwent six months after the first treatment with cytostatic drugs. Before the transplantation, S. epidermidis was again isolated from three different blood cultures through the central venous catheter, although the patient had no fever. Isolate E4 was the first of these and was recovered nine days before ES. The antibiograms of these isolates were identical to that of E3. The catheter was removed and a new one inserted while vancomycin was being given. Typing methods Antibiogram. Standardized disc diffusion susceptibility formed on Isosensitest agar (Unipath Ltd, Basingstoke, biotics tested are shown in Table I.
tests’ were perUK). The anti-
34
G. Hedin
Biotyping. Biotyping was performed with a commercial kit for species t.yping of staphylococci, ‘API Staph’ (API Bio Merieux SA. Lyon, France).
Slimeproduction.
Slime production was determined by the tube adherence test described by Christensen et ~1.” Slime production was recorded as negative (-), weak (+), moderate (2+) or strong (3+) according to the density of the adherent biofilm.
Plasmid
profile. Plasmids lysis method.” They were and stained with ethidium and photographed. Phage molecular size marker. PFGE
were isolated by a modification separated on a 0.7% agarose gel bromide before being visualized h DNA digested with Hind111
of the alkaline at 80 V for 1 h with UV light was used as a
of chromosomal DNA. Bacterial colonies from blood agar were suspended in 0.5 ml PIV buffer (1 M NaCl, 10 mM Tris-HCl, pH 8.0) containing 250 ltg lysostaphin (Sigma, St. Louis, USA) and 2 mg lysozyme (Sigma). An equal volume of TE buffer ( 10 mM Tris-HCl, 0.1 mM EDTA, pH 7.5) containing 2% agarose (Sigma Type VII: low gelling temperature) at 55°C was then added. The mixture was immediately poured into the slots of a plastic mould (2 x 5 x 10 mm), allowed to solidify at room temperature and incubated at 37°C for 2 h. The agarose plugs were transferred to tubes each containing 3 mL lysis buffer (6 mM Tris-HCl, 10 mM EDTA, 0.2% deoxycholate, 0.5 lauroyl sarcosine, pH 8.0) with RNase (50 l.rg/ mL)(Boechringer GmbH, Mannheim, Germany) and incubated at 37°C for 2 h. The lysis buffer was replaced by proteolysis buffer (0.5 M EDTA, 1% lauroyl sarcosine) containing 1 mg/mL proteinase K (Boehringer). After overnight incubation at 50°C the agarose plugs were washed with distilled water and then with 2 mL TE buffer to which 30 PL of a 100 mM solution of phenylmethylsuphonyl fluoride in isopropanol was added. After incubation at room temperature for 1 h, three washing steps followed, first with distilled water, then two washes with TE buffer. The agarose plus were stored in TE buffer at 4°C until the next day. The restriction enzyme SmaI (Gibco BRL, Gaithersburg, USA) was used, each agar plug being incubated at room temperature for 4 h in a tube containing 100 U of the enzyme in 100 FL buffer. Small fragments of each agarose plug were placed into the slots of a 1% agarose gel and the slots sealed with 1% molten agarose (Sigma type II. Medium EEO). The DNA restriction fragments were separated by PFGE in a CHEF-DR II system (Bio-Rad Laboratories, Nazareth, Belgium). Electrophoresis was carried out at 180 V in 0.5 x TBE buffer (0.045 M Tris-borate, 0.001 M EDTA) with alternating pulses at a 120” angle in a 7-12 s pulse time gradient for 11 h and then a 20-40 s gradient for 13 h at 14°C. Pharmacia h DNA-PFGE marker (Pharmacia, Uppsala, Sweden) was used as a molecular size marker. The gels were stained with ethidium bromide and photographed under UV light.
S. epidermidis Table
I. Antibiograms
of 20 isolates
typing
of Staphylococcus
Antibiotic
Benzylpenicillin Methicillin Co-trimoxazole Doxycycline Erythromycin Clindamycin Chloramphenicol Rifampicin Vancomycin Ciprofloxacin Fusidic acid Gentamicin Netilmicin Tobramycin Amikacin
methods
35
epidermidis Isolate
from
Bl-B2
Cl-C7
Dl,D3
D2,D4
R R R
R R R S
R R R s
:: R S
:: R S R
E R
: S
: R
ii S R
: S
:
:
: S
: ::
: R R
: R R
:: R
patients
No.
Al-A2
ii S
jive
:: R R
El-E2
E3-E5
R
R R
:: R S
:: z
:: S z R R R R R
: S S R S
::
: R R
: R R
: S
Level of resistance to methicillin. The ‘Bioscreen’ analysing system (Labsystems, Helsinki, Finland) was used for continuous measurement of bacterial growth in broth containing methicillin.12 Bacterial colonies from the overnight growth on blood agar was suspended in 2 mL of phosphatebuffered saline and the cell density was adjusted to 10’ cfu/mL. This suspension was serially diluted lo-fold in Mueller Hinton broth (Difco Laboratories, Detroit, USA) supplemented with 2% NaCl, calcium, 50 mg/ L, magnesium, 25 mg/L, and methicillin, 16 mg/L. The dilutions (0.3 mL volumes) were transferred to the wells of specially designed microplates supplied by the manufacturer (Labsystems). For each strain, five different inocula were tested: 103, 104, 1OS,1O6and 1O7cfu, each generating a different growth curve in the ‘Bioscreen’ system. Growth was observed first in the wells with the highest inoculum sizes and later in the wells with the lower inoculum sizes. Strains expressing heterogeneous resistance to methicillin showed growth within 24 h only in the wells with the third or fourth highest inoculum sizes. The growth curves of different isolates were judged by the naked eye, and those isolates with similar curves were grouped together. This was repeated at least three times for each strain. Results Three of five patients (A, B and C) each yielded isolates of S. epidermidis with identical antibiograms. Of the remaining two patients, patient D harboured isolates with two different antibiograms and so did patient E. Table I shows that in total seven antibiogram patterns were identified and variability in resistance was most often seen with chloramphenicol and
36 Table
G. Hedin II.
Patient isolate Al ,A2 Bl.B2
a-c7 Dl D3 D2,D4 El E2 E3 E4,45
Summary
of the results
Antibiogram profile
of different
Slime production
A
1+
:
2+ 3+
E
2
E F F
i::
E
1+
methods Plasmid pattern
for
strain
characterization
PFGE profile I I II III III IV I I V VI
2+ 2+
Methicillin sensitivity* 1 : ;
4 4 2 ;
*See text.
fusidic acid. All isolates were uniformly resistant to amikacin and tobramycin but resistance to gentamicin and netilmicin was variable. Table II summarizes the results of other characteristics tested for the isolates. All isolates were uniform in biotype and all produced slime but to varying degrees. Slime production was strongest for all isolates from patient C and two from patient D. Seven plasmid patterns and six DNA macrorestriction profiles were found for all isolates. Only four levels of methicillin resistance were identified. Within each patient, the combination of methods showed that those from patients A, B and C were uniform indicating infection by a single strain unique for each patient. However, three strain phenotypes were found among isolates from patient D. Isolates Dl and D3 were identical in all respects except for the presence in Dl of a plasmid of ~2000 bps, which was present also in D2 and D4. The latter were indistinguishable from each other but differed from other isolates although some shared characters such as plasmid content, response to slime production test and level of resistance to methicillin were observed. The isolates from patient E showed the greatest heterogeneity in typing markers. These were characterized by two plasmid patterns, three PFGE profiles, and two responses to methicillin. Resistance to methicillin was expressed by all cells in the population in isolate El, whereas in the remaining isolates from this patient resistance was expressed by only a minority of cells. Discussion
In this study different typing methods were used to answer a common clinical question of whether isolates of S. epidermidis from one and the same patient are identical. Twenty S. epidermidis isolates from five different patients were selected on the basis that they had identical or closely similar antibiograms.
S. epidermidis
typing
methods
37
(b)
(a)
Figure 1. Plasmid patterns of Staphylococcus epidermidis strains isolated from patients A, B, D and E. (a) Lane 1: molecular size marker (Hind111 digest of h DNA); lane 2: Al; lane 3: A2; lane 4: Bl; lane 5: B2; lane 6: Dl; lane 7: D2; lane 8: D3; lane 9: D4. (b) Lane 1: molecular size marker; lane 2: El; lane 3: E2; lane 4: E3; lane 5: E4; lane 6: ES.
1
2
3
4
5
6
Figure 2. Plasmid patterns of Staphylococcus epidermidis Lane 1: molecular size marker (Hind111 digest of h DNA); C3; lane 5: C4; lane 6: C5; lane 7: C6; lane 8: C7.
7
strains lane
8
isolated from patient C. 2: Cl; lane 3: C2; lane 4:
It is known that the characteristics of a strain, even those commonly used for epidemiologic typing, may become modified over time. A given strain of S. epidermidis may acquire or lose plasmids and/or antibiotic resistance genes as observed in studies of strains isolated from the same patient on different occasions.13-15 PFGE patterns of these strains may undergo also slight modifications with time.16 Despite these reservations, the antibiogram and plasmid content of some strains of S. epidermidis may remain stable for several months.17 The isolates of S. epidermidis used in this study were recovered from the patients at time intervals of three, four,
G. Hedin 12345678
9 10 11 12 13 14
Figure 3. Pulsed-field gel electrophoresis of SmaI digests of chromosomal DNA. All different banding patterns for 20 isolates of Staphylococcus epidermidis are shown. As the patterns of all the C strains were identical, only that of C7 is shown and as the pattern of Al was identical to that of A2, Bl, B2, El and E2, only those of Al, El and E2 are shown. Lane 1: C7; Lane 2: molecular size marker (Pharmacia h DNA-PFGE marker: X1857 Sam 7 ligated concatemers, sizes 48.5, 97.0, 145.5 kb . . . etc.); lane 3: Dl; lane 4: D2; lane 5: D3; lane 6: D4; lane 7: molecular size marker; lane 8: Al; lane 9: El; lane 10: E2; lane 11: E3; lane 12: E4; lane 13: ES; lane 14: molecular size marker.
24 and 19 days for patients A-D, and with a time interval of six months in patient E. Overall the 20 isolates exhibited a single biotype, seven antibiograms, seven plasmid profiles, six PFGE patterns and four levels of resistance to methicillin. Isolates identical according to one typing method were not invariably grouped together by another method and so the combination of results identified 10 groups of strains. Biotyping has a low discriminatory power. The majority, 70% or more of blood culture isolates of coagulase-negative staphylococci, belong to the species S. epidermidis, most of which have the same biotype.” Nevertheless biotyping should be included in a typing scheme to make identification of an unusual biotype possible. Both homogeneous and heterogeneous expression of methicillin resistance was displayed by the isolates. Those that expressed heterogeneous resistance produced different growth curves. Four types of growth curve response to methicillin were observed in the present study, which may be compared with the four arbitrary expression classes proposed by Tomasz et aZ.* for strains of staphylococci. The two isolates from patient A were identical by all the methods used, as were those from patient B and patient C. The A and B isolates came
S. e$idermidis
typing
0
39
methods
12
24
0.8
I 0
12
24 Hours
Figure 4. Growth curves in broth containing methicillin for the 20 isolates of S. epidermidis. The growth curves for each of five inoculum sizes are shown in the same figure. The growth curves of strains isolated from the same patient considered as identical are superimposed on one another in the figures. (a) Al and A2. (b) Bl and B2. (c) D2 and D4. (d) Dl and D3. (e) El. (f) E2-ES. (g) Cl-C7.
from the same department and had identical PFGE patterns and levels of resistance to methicillin. The A isolates were, however, susceptible to fusidic acid, whereas the B isolates were resistant, the latter also contained an additional plasmid and produced slightly more slime. The PFGE pattern
40
G. Hedin
of the A and B strains was also identical to that of the first and second strains from patient E. The unreliability of plasmid profiling as a sole method for epidemiologic typing was illustrated by the results on isolates from patient D. The antibiograms, PFGE patterns, slime production and level of resistance to methicillin all suggested that Dl and D2 were identical to each other and different from D3 and D4, which in turn were identical to each other. The plasmid profiles of the strains, however, suggested that Dl, D2 and D4 were identical, while D3, where a small plasmid was missing, was different. The strains isolated from patient E were interesting also because six months had elapsed between the isolation of the first and last strain. Genetic changes among the strains were not therefore completely unexpected. The typing results suggested that a continuous genetic change had occurred within a number of strains with a common origin, i.e., a clone. The difference between the first two strains isolated from patient E was subtle. These strains had identical antibiograms, plasmid and PFGE patterns, and slime production. The only difference between them was that in El, methicillin resistance was expressed homogeneously, whereas in E2 it was expressed heterogeneously. This was the only case in the present study where the level of resistance to methicillin was found to be unstable in viva. The third strain, E3, was isolated one month later. This strain had acquired resistance to fusidic acid and the PFGE pattern also differed slightly from that of El and E2. Plasmid pattern and slime production, however, were identical to those of the earlier strains. According to the guidelines for strain typing by PFGE recently proposed by Tenover et ~1.‘~ the E3 strain was classified as possibly related to El and E2. The restriction pattern of El and E2 contains two bands (sizes approximately 390 and 110 kb), which are missing in the restriction pattern of E3, where instead two other bands (sizes approximately 240 and 260 kb) are present. In fact, a change in the position of only one restriction site may explain the observed differences. The last two strains, E4 and ES, were isolated five months later. They had the same antibiogram as the earlier strains, but an additional plasmid was seen, the slime production was slightly less profuse, and the PFGE pattern was also slightly modified. The level of resistance to methicillin was identical in strains E2-ES. In summary, examples were seen in this study where one typing method indicated a difference between two isolates, though a relationship was clearly indicated by the other methods, i.e., the strains belonged to the same clone. Accordingly, the plasmid profile was observed to change within a clone of strains from a patient, as was the PFGE pattern and the level of resistance to methicillin within a clone from a second patient. Identical PFGE patterns were found also in isolates that were not epidemiologically related. Thus no single method can provide the answer to the question of the relatedness of clinical isolates. Ultimately, the most reliable result will be obtained if different methods are used in combination. In the present
5’. epidermidis
typing
methods
41
study, identical antibiograms among isolates of S. epidermidis from the same patient indicated always that the strains were closely related, but this may not always be the case.4 It seems, therefore, that in clinical practice the use of antibiograms for epidemiologic typing is a natural first step. The level of resistance to methicillin is stable in viva in strains of S. epidermidis, as shown in this study, and therefore, can be used as an additional aid to identifying strains. The pulsed-field gel electrophoresis Copenhagen, Denmark, by the author. Seruminstitut, for her assistance.
studies were I am grateful
performed at Statens to Dr B. Thamdrup-Rosdahl,
Seruminstitut, Statens
References 1. Pfaller MA, Herwaldt LA. Laboratory, clinical, and epidemiological aspects of coagulase-negative staphylococci. Clin Micvobiol Rev 1988; 1: 281-299. 2. Archer GL. Coagulase-negative staphylococci in blood cultures: the clinician’s dilemma. Infect Control 1985; 6: 477478. 3. Ludlam HA, Noble WC, Marples RR, Philips I. The -evaluation of a typing scheme for coagulase-negative staphylococci suitable for epidemiological studies. J Med Microbiol 1989; 30: 161-165. 4. Khatib R, Reiderer KM, Clark JF, Khatib S, Briski LE, Wilson FM. Coagulasenegative staphylococci in multiple blood cultures: strain relatedness and determinants of same-strain bacteremia. J Clin Microbial 1995; 33: 816-820. 5. Chambers HF. Coagulase-negative staphylococci resistant to p-lactam antibiotics in vivo produce penicillin-binding protein 2a. Antimicrob Agents Chemother 1987; 31: 1919-1924. 6. Chambers HF. Methicillin-resistant staphylococci. Clin Microbial Rev 1988; 1: 173-186. BJ, Tomasz A. Expression of methicillin resistance in heterogeneous strains 7. Hartman of Staahvlococcus aureus. Antimicrob Agents Chemother 1986: 29: 85-92. 8. Tomasz *A, Nachman S, Leaf H. Stableclasses of phenotypid expression in methicillinAntimicrob Agents Chemother 1991; 35: resistant clinical isolates of staphylococci. 124-l 29. 9. Swedish Reference Group for Antibiotics. The Swedish SIR-system for susceptibility testing of bacteria. SGRA 1991. Stockholm. 10. Christensen GD, Andrew-Simpson W, Bisno AL, Beachey EH. Adherence of slimeproducing strains of Staphylococcus epidermidis to smooth surfaces. Infect Immun 1982; 37: 318-326. HC, Doly J. A rapid alkaline extraction procedure for screening recombinant 11. Birnboim plasmid DNA. Nucleic Acids Res 1979; 7: 1513-1523. G, Hambraeus A. Screening tests for the detection of methicillin resistance in 12. Hedin Staphylococcus epidermidis. J Antimicrob Chemother 1991; 28: 681-694. 13. Etienne J, Renaud F, Bes M et al. Instability of characteristics amongst coagulasenegative staphylococci causing endocarditis. J Med Microbial 1990; 32: 115-l 22. H, Jask D, Hammerberg 0. Evaluation of restriction endo14. Bialkowska-Habrzanska nuclease fingerprinting of chromosomal DNA and plasmid profile analysis for characterization of multiresistant coagulase-negative staphylococci in bacteremic neonates. J Clin Microbial 1990; 28: 269-275. PA, Plorde JJ, Gordon KP et al. Instability of antibiotic resistance in a strain 15. Mickelsen of Staphylococcus epidermidis isolated from an outbreak of prosthetic valve endocarditis. J Infect Dis 1985; 152: 50-58. 16. Goering RV, Winters MA. Rapid method for epidemiological evaluation of Grampositive cocci by field inversion gel electrophoresis. J Clin Microbial 1992; 30: 577-580. G, Hambraeus A. Enhanced ability to colonize the skin: a possible explanation 17. Hedin for the epidemic spread of certain strains of Staphylococcus epidermidis. J. Hosp Infect 1993; 25: 251-264.
42 18.
G. Hedin Hedin G. Comparison of coagulase-negative
of genotypic staphylococcal
and phenotypic isolates from
methods for blood cultures.
species
APMIS
identification
1994; 102:
855-864. 19.
Tenover patterns
FC, Arbeit, RD, Goering, produced by pulsed-field
J Clin Microbial.
RV et al. Interpreting gel electrophoresis:
1995; 33: 2233-2239.
chromosomal DNA restriction criteria for bacterial strain typing.