Molecular Tracking of Coagulase-Negative Staphylococcal Isolates From Catheter-Related Infections

Three molecular typing methods (pulsed-field electrophoresis, localization of the mecA gene, and probing the vicinity of mec) have been used for the characterization of 40 catheter-related isolates of coagulase-negative staphylococci (CNS) in 14 patients admitted to the same hospital. The 40 isolates yielded 14 different Smal banding patterns and corresponding unique localizations of mecA, each associated with a unique Clal mecA polymorph. In 6 of the 14 patients the contaminated skin at the catheter entry site was the source of 4 local infections and 2 cases of bacteremia. A contaminated hub was the origin of 2 local infections and 4 cases of bacteremia in 6 more patients. The remaining 2 patients had positive cultures from both skin and catheter hub. In each bacteremic patient, the CNS recovered from catheter-related sites (tip, skin, and/or hub) and the CNS recovered from blood were identical, but each of these matching isolates was unique to the particular patient, indicating a low rate of cross-infection from patient to patient. Although classical methods for typing CNS (e.g., biotype and antibiotype) are readily available for most hospital laboratories, they have limitations concerning reproducibility and discriminatory power. Molecular epidemiologic techniques can provide powerful support to traditional techniques in determining the etiologic role of CNS in the disease process.


INTRODUCTION
The most common pathogens associated with catheter-re- lated infections (CRIs) are coagulase-negative staphylococci (CNS), specifically Staphylococcus epidermidis.The in- crease in the use of indwelling devices, along with changes in the patient population, hospital environment, medical practice, and antibiotic policy, have turned these microorganisms into the leading cause of hospital-acquired bacteremia.I4-26 '30 The pathogenesis of infections associated with intravascular catheters is closely related to the cutaneous surface flora.5][16][17][18][19][20][21][22][23][24][25][26] The two leading routes of catheter contamination are invasion of the catheter external surface by microorganisms present at the insertion site14-24 and inner mi- gration of bacteria to the catheter tip through the lumen from a contaminated hub.19-27-30 Exogenous microorganisms can be in- oculated into the catheter wound or into the hub by contami- nated hands of hospital personnel or contaminated antiseptics.14Endogenous microorganisms from other body sites such as the respiratory and gastrointestinal tracts can colonize these areas as well.Hematogenous seeding from a septic distant focus and contamination of the infúsate can cause CRI as well.14But none of these routes rivals the importance of the patient's own cuta- neous flora contaminating by itself the catheter insertion site or even being transferred to the hub after contacting the patient's skin.14CNS also serve as reservoirs for antibiotic resistance genes3 that may be transferred to other gram-positive organisms.This could be the case of the mecA gene.The structural gene mecA, the central genetic determinant of methicillin resistance, has been cloned23-32 and found essentially identical in all species of staphylococci, both coagulase-positive and coagulase-negative, that have the characteristic methicillin-resistance phenotype.2-28-29-32The variability associated with the mecA region has been used as a valuable typing tool among methicillin-re- sistant Staphylococcus aureus. 11,12fforts to identify the source of CRIs and modes of their This work was partially presented at the 7th European Congress of Clinical Microbiology and Infectious Diseases, March 1995, Vienna, Austria, (abstract no.1560).
2The Rockefeller University, New York, New York 10021, and Instituto de Tecnología Química e Biológica (ITQB), Oeiras, Portugal.transmission within the hospital have been hampered by the lack of sufficiently discriminatory typing systems that could distinguish between strains colonizing skin and those causing invasive disease.17'26 Traditional epidemiologic analysis needs to be substantiated by more sophisticated molecular tech- niques,1030 which have been reported as extremely useful in the global evaluation of hospital reservoir of CNS9'13'15 '17-33 and in the particular case of CRIs.30The present study evaluates the use of 3 molecular typing methods to characterize 40 CNS isolates associated with infected central venous catheters in 14 patients.Special emphasis is placed on the use of an S. aureus mecA gene probe in the intraspecific typing of methicillin-resistant CNS.

Bacterial isolates
Forty catheter-related CNS isolates were collected from 14 inpatients admitted in a 1000-bed hospital in Barcelona, Spain.Seven patients were admitted in a general intensive care unit (cases 1 to 7), and 7 were from a general surgery unit (cases 8 to 14).All patients had central venous, catheters; patients 8 to 14 were receiving total parenteral nutrition.Catheters were re- moved when suspected of involvement with infection.The signs considered for CRI were any of the following: unexplained fever, localized inflammation, and/or pain on the vascular line and/or on the skin at the catheter entry site.The catheter in- travascular segment (3-4 cm) was sent for microbiologie analy- sis.Swabs from the hub and skin surrounding the catheter in- sertion site, taken before catheter withdrawal, were also cultured.Blood cultures, obtained by a distant vein puncture, were performed as well.

Microbiological methods
Catheter cultures were performed by the semiquantitative (SQ) method22 and by the modified quantitative (MQ) method described by Linares et al.19In the SQ method the catheter tip was transferred to the surface of a blood agar plate and rolled back and forth across the agar surface.After this, the MQ method was carried out by flushing the catheter lumen with 1 ml tripticase soy broth (Becton-Dickinson Microbiology Systems, Cockeysville, MD), which was then diluted 10-fold, and 0.1 ml of each dilution was plated on blood agar.Colonies were counted after 48 h of incubation.When 15 or more colonyforming units (CFU) grew on the SQ culture, the catheter was considered infected; if less than 15 grew, the catheter was con- sidered colonized.For the MQ method, more than 103 CFU/ml was necessary to consider the catheter infected.
CNS recovered from the samples were identified to the species level by an automatized system, Microscan (DADE International, West Sacramento, CA).Antibiotic susceptibility was determined by the disk diffusion and the microdilution methods, according to National Committee for Clinical Laboratory Standards criteria. 25The following antibiotics were tested: penicillin, oxacillin, cephalothin, vancomycin, ery- thromycin, clindamycin, gentamicin, trimethoprim-sul- famethoxazole, fucsidic acid, and ciprofloxacin.

Molecular analysis
The purification of whole genomic DNA was done accord- ing to methods established for S. aureus.11Three molecular chromosomal typing methods were employed: (1) pulsed-field gel electrophoresis (PFGE) after DNA restriction with Smal and separation of fragments in a contour-clamped homogeneous electric field (CHEF-DRII) apparatus (Bio-Rad, Richmond, CA), with running conditions of 200 V and pulses ranging from 1 to 30 sec during 23 h at 14°C; (2) Smal fragments separated by PFGE were hybridized with the mecA-specific DNA probe1 ' for the localization of the gene on the chromosomal fragments; and (3) digestion of DNA with Clal followed by hybridization with a mecA-specific DNA probe.DNA fragments were trans- ferred to Hybond-N+ (Amersham, Buckinghamshire, UK) ny- lon membranes" and hybridized with a nonradioactively la- beled probe (ECL, Amersham).
Analysis and interpretation of data DNA in PFGE was visualized and photographed in ultravi- olet light after the gels were stained with ethidium bromide.PFGE patterns with more than 3-band differences were assigned to distinct clonal types.31Original autoradiograms following hybridization with the mecA probe were analyzed.Differences in a single band were considered as different mecA polymorphisms.The discriminatory power (DP) percentage in typing CNS was calculated as follows: [number of types/total number of strains] X 100 for a given typing technique.20

Microbiological results
The 40 catheter-related CNS and their sites of isolation from 14 patients are shown in Table 1. S. epidermidis was recovered from 12 of the 14 patients, and the total number of isolates of 5. epidermidis was 33.Staphylococcus auricularis (2 isolates) was recovered from patient 5 and Staphylococcus saprophyticus (3 isolates) from patient 11.Staphylococcus haemolyticus was re- covered from mixed infections of patient 6.All 14 catheter tips were found infected by the SQ culture (>15 CFU) and/or the MQ culture (>103 CFU/ml).Of the initial CRI in 14 patients, 6 patients had catheter insertion site (skin) cultures that were pos- itive for CNS (cases 2, 3, 5 to 7, and 9); 6 had hub cultures that were positive (cases 1 and 10 to 14); and the remaining 2 pa- tients (cases 4 and 8) yielded CNS from both skin and hub cul- tures.Two patients (cases 6 and 7) had the tip infected with 2 different CNS, both of which were also recovered from the skin in case 6.Seven patients had catheter-related bacteremia, and the isolate recovered from blood was added to the study.Thus there were 7 cases of catheter-related bacteremia (4 likely originated from the contaminated hub, 2 from the skin insertion site, and 1 with a combined origin) and 7 local catheter infections that did not progress to bacteremia, presumably because the contaminated devices were removed in time.
Phenotypic traits of CNS-related strains All CNS were resistant to oxacillin and at least 3 additional antimicrobials.All isolates retained susceptibility to van- aNC, negative culture.bNH, not homologous with mecA probe.
The CNS recovered from all different catheter-related sites showed the same phenotypic traits (biotype and antibiotype) in each patient (Table 1).
Molecular traits of CNS-related strains DNA fingerprinting of the 33 isolates of S. epidermidis (iso- lated from 12 patients) yielded the following results (Table 1): PFGE after Smal digestion of chromosomal DNA resolved them into 11 different banding patterns (A to F, I to K, M and N), each one with a characteristic location of the mecA gene and each carrying a unique mecA polymorph as well.For instance, the two CNS isolates in case 1 showed a unique PFGE pattern (Fig. I, A); the mecA gene probe hybridized with a 48-kb Smal fragment (Fig. 1, B); and mecA polymorph 1 was present (Fig. 2, A).In contrast, the 3 CNS isolates of case 2 shared a com- mon PFGE pattern B (Fig. 1, A) and a Clal mecA polymorph 2 (Fig. 2, A).In this group of isolates the mecA probe hybridized with a 90-kb Smal fragment (Fig. 1, B).Similarly, the remain- ing 28 S. epidermidis isolates each had unique PFGE patterns, carried mecA polymorphs, and showed mecA-hybridizing fragments of unique molecular size.
The 2 pairs of S. auricularis and S. haemolyticus isolates (cases 5 and 6) belonged to 2 additional and unique PFGE types named G and H, each with its characteristic mecA location and carrying 2 corresponding, unique mecA polymorphs (polymorphs 7 and 8).Three S. saprophyticus isolates from case 11 also had a single PFGE pattern but did not show any homol- ogy with the mecA fragment used as probe; thus no mecA hy- bridizing pattern was found.All together, of 40 CNS isolates, PFGE yielded 14 different patterns (DP, 35%) and Clal mecA hybridization, 13 types (DP, 32.5%).
As shown in Table 1, in every single patient a unique and identical CNS clone (as defined by the molecular, biochemi- cal, and antibiotic resistance profiles) could be recovered from all different catheter-related sites.Furthermore, in the 7 bac- teremic patients bacteria recovered from blood cultures were identical to the bacteria isolated at the catheter sites of each particular patient.We conclude that extraluminal contamina- tion of the catheter's external surface from the skin at the catheter entry site has been the most likely origin of 4 local infections (cases 3, 5, 6, and 7) and 2 cases of bacteremia (cases 2 and 9).The inner catheter surface was most likely contaminated from a contaminated hub in 6 patients and was the origin of 2 local infections (cases 1 and 10) and 4 bac- teremias (cases 11, 12, 13, and 14).In cases 4 and 8, positive cultures were obtained from both skin and hub.Presumably, in these cases hubs were contaminated by inoculating the pa- tient's own skin flora in the subsequent manipulation of the central line.

DISCUSSION
The pathogenesis of central catheter infections and the role of CNS have been properly established since the early 1980s. 19,21nce then, every scientific review of the field remarks the need for more reliable tools to conclusively prove the relatedness of the isolated strains.14-26'30Methods that still are recommended for typing CNS are the biochemical profile and the antimicrobial sus- FIG. 1. Chromosomal backgrounds and localization of mecA in isolates of coagulase-negative streptococci (CNS). (A) Chro- mosomal DNAs of 12 CNS isolates from 5 patients were re- stricted with Smal.Five unique patterns were detected by pulsed-field gel electrophoresis (PFGE).Among the 10 Staphylococcus epidermidis isolates, patterns were shared by isolates 1 and 2 (pattern A), 3-5 (pattern B), 6 and 7 (pattern C), and 8-10 (pattern D).Lanes 11 and 12 contain Staphylococcus haemolyticus pattern H. (B) The same gel shown was hybridized with the mecA DNA probe to localize the gene to various Smal fragments.A ladder and A l.m.w. (low molecular weight) are lambda ladder PFG markers (New England BioLabs, Beverly, MA).ful in analyzing CNS from distinct catheter-related sites when a single patient was considered.Considered separately, neither bio- type nor antibiotype was adequate in tracing routes of CNS trans- mission among different patients (i.e., isolates with the same bio- type or antibiotype belonged to different clonal types, as confirmed by molecular methods).However, if both phenotypic typing techniques were considered together, their DP value rose to 37%, a percentage even higher than that obtained with the mol- ecular tools applied here (35%).It should be mentioned at this point that the automated system used in our clinical laboratory for CNS speciation contained 26 biochemical tests and 21 antibiotics.
Many sophisticated and molecular techniques have been ap- plied recently for studying the epidemiology of CNS infections.Some examples are multilocus enzyme electrophoresis,30 plas- mid DNA analysis,4-10-24 restriction endonuclease fingerprinting of the chromosomal DNA4 and hybridization with specific nu- cleic acid probes,3-5-9 PFGE after genomic DNA macrorestric- tion,8-10 and polymerase chain reaction-based typing techniques.6 In the study described here, we used 3 molecular typing tools that have already demonstrated good discriminatory power in the analyses of methicillin-resistant S. aureus (MRSA) hospital out- breaks.5"Two of these typing tools were based on determining mecA polymorphism after digestion of chromosomal DNA with Smal or CM.All but 1 of the CNS isolates studied here were nosocomially acquired, and all of them were multidrug-resistant able on a timely and affordable basis for most hospital laborato- ries.However, they have recognized limitations.They do not al- ways give reproducible results10; unique patterns may result from local conditions, such as antibiotic pressure, and may not be sta- ble over time or geographic distance.Appearance of unique bio- chemical or susceptibility patterns may serve as a preliminary marker that will need to be confirmed by complementary tech- niques.In our study biotype or antibiotype was shown to be use- FIG. 2. Chromosomal DNAs from CNS were restricted with Clal and hybridized with the mecA probe; mecA polymorphs were assigned. (A) The same isolates of S. epidermidis as in Figure 1 are shown (5.haemolyticus is not shown).Isolates from cases 1 to 4 showed polymorphs 1, 2, 3, and 4, respec- tively. (B) Case 5: Staphylococcus auricularis, polymorphism type 7. Case 6: S. haemolyticus (lanes 3 and 5), polymorphism type 8; and S. epidermidis (lanes 4 and 6), polymorphism type 5. Case 7: S. epidermidis (lane 7), type 1; and S. epidermidis (lanes 8 and 9), type 6.
and methicillin-resistant, allowing the use of the mecA gene as a molecular marker.This would remain the case for the majority of nosocomial CNS infections, since in our nosocomial environ- ment, about 70% of CNS are oxacillin resistant.A similar figure was reported by the National Nosocomial Infection Survey for CNS isolates in the United States, in 1989, with the most fre- quent coagulase-negative species being S. epidermidis.1-15 Our findings indicate that the DP of the mecA probe may be higher than that reported for ribotyping16 among CNS.Also, it showed higher DP among CNS than among MRSA isolates.Only relative few mecA polymorphs have been identified among sev- eral hundred MRSA strains collected from a wide variety of ge- ographic sources,18 and the mecA hybridization patterns of hos- pital outbreaks of MRSA disease are usually associated with a dominant MRSA clone carrying a single mecA polymorph.1112 This is in sharp contrast to the observations described in this com- munication: the CNS isolates collected at a single hospital site within a relatively short period showed a wide range of variation in the vicinity and localization of the mec gene, and also in the general chromosomal backgrounds of the isolates.This fact has allowed an unambiguous determination of the origins and trans- mission routes of the CNS isolates involved with the 14 CRIs described here.
Previous studies have demonstrated that some CNS can be- come endemic over long periods and can be transmitted among patients by hospital personnel.915In our particular set of CNS infections, there were only 2 cases of potential cross-infection: clone A::l (PFGE::mecA polymorph) from case 1 (isolated from skin and catheter tip), which was also found in case 7 (from catheter tip) 2 months later; and clone I::9, which was found in case 8, and 3 months later in case 12.
The introduction of methods such as those used in this study should generally improve the management of nosocomial in- fections by CNS strains.CNS infections are notoriously diffi- cult to treat, both because of the patient population affected (long-term hospitalized patients, often with severe underlying disease) and because of the multidrug resistance expressed by these bacteria.CNS are also the most common contaminants in the hospital environment, and molecular techniques offer the high discriminatory power essential for distinguishing contam- inants from the true etiologic agents of disease.

7
Both biotype and antibiotype are readily avail-

Table 1 .
CNS Clinical Sources and Clonal Types