Emergence and Spread in French Hospitals of Methicillin-Resistant Staphylococcus aureus with Increasing Susceptibility to Gentamicin and Other Antibiotics

Abstract

Oxacillin (methicillin) resistance in methicillin-resistant Staphylococcus aureus (MRSA) is associated with an increased incidence of resistance to other antibiotics, which has increased since it was first reported in 1969. In 1992 a new phenotype of MRSA arose in France; this was characterized by a heterogeneous expression of resistance to oxacillin and susceptibility to various antibiotics, including gentamicin but also tetracycline, minocycline, lincomycin, pristinamycin, co-trimoxazole, rifampin, and fusidic acid. In French hospitals a longitudinal nationwide surveillance of antibiotic resistance in S. aureus has allowed for the detection of changes in antibiotic susceptibility profiles. Seven French clinical laboratories (six from the mainland and one from the West Indies) reported the results of susceptibility testing of 57,347 S. aureus strains isolated in their institutes between 1992 and 1998. Over a 7-year period the incidence of isolation of gentamicin-susceptible MRSA (GS-MRSA) strains has steadily increased to represent, in 1998, 46.8 to 94.4% of the MRSA strains, irrespective of the overall incidence of MRSA. Two predominant types recognized by pulsed-field gel electrophoresis (PFGE) accounted for the majority of the GS-MRSA in different mainland hospitals, both differing from the predominant type observed in the French West Indies. Some GS-MRSA and gentamicin-resistant MRSA (GR-MRSA) strains had closely related PFGE profiles, and hybridization studies confirmed the lack in GS-MRSA of the aac6′-aph2" gene, which confers resistance to all aminoglycosides, with conservation of the ant4′ gene, which confers resistance to kanamycin, tobramycin, and amikacin. Thus, it is likely that certain GS-MRSA strains could have emerged from GR-MRSA strains by excision or deletion of the aac6′-aph2" gene.

In many countries oxacillin (methicillin)-resistant Staphylococcus aureus (MRSA) has become a significant nosocomial pathogen. MRSA was first reported in the United Kingdom in 1961, soon after the introduction of methicillin, and by the mid-1970s had become endemic in many countries (26). Some strains of MRSA have been designated epidemic strains; these are associated with a higher prevalence and have been shown to have spread within hospitals, between hospitals, and between countries (1, 10, 16, 17, 21). The first MRSA isolates expressed so-called heterogeneous phenotypic resistance to oxacillin, meaning that the oxacillin MICs for only subpopulations of isolates are high. Progressively, the heterogeneous oxacillin-resistant phenotype was replaced by the homogeneous oxacillin-resistant phenotype, which is characterized by the expression of oxacillin resistance by all populations. Initially, early isolates were also resistant to various other drugs, including penicillin, tetracycline, and, usually, streptomycin (some strains were also resistant to erythromycin, lincomycin, neomycin, kanamycin, and novobiocin). In 1969, the first clinical gentamicin-resistant MRSA (GR-MRSA) strain was isolated (11), and by the 1980s GR-MRSA had become epidemic in Australia, the United States, and Europe (5). Such GR-MRSA strains were usually resistant to a broad number of other antibiotics, including trimethoprim and, more recently, ciprofloxacin and mupirocin. In addition to increasing multi-antibiotic drug resistance, the overall incidence of MRSA isolation has gradually increased in many countries to present levels of around 30% in Spain, France, and Italy (26) and up to 54% in Japan (14). The emergence of new epidemic MRSA strains more susceptible to antibiotics has been recently reported by two French hospitals (2, 12). These strains were characterized mainly by the unexpected reappearance of heterogeneous resistance to oxacillin, susceptibility to gentamicin, and variable resistance to macrolides, lincosamides, and streptogramin type B antibiotics; they remained resistant to tobramycin, which was associated with the presence of the aminoglycoside nucleotidyltransferase ANT(4′) (2, 12). A marked decrease in the use of gentamicin was suspected to be a factor contributing to the emergence of gentamicin-susceptible MRSA (GS-MRSA) from predominantly GR-MRSA populations (2, 12).

The aim of the present study was to investigate whether the previously reported evolution of antibiotic resistance of MRSA in two hospitals in the Paris area could be relevant at the nationwide level, by collecting antibiotic susceptibility data from seven French clinical laboratories between 1992 and 1998. Subsets of isolates were retrieved and studied in more detail in an attempt to better understand the molecular basis of increasing gentamicin susceptibility.

MATERIALS AND METHODS

Source of bacterial strains and analysis of antibiotic susceptibility.

Clinical laboratories from seven hospitals dispersed throughout French territory (hospitals A, B, and C in Paris; hospital D in Lille [northern France]; hospital E in Lyon [central France]; hospital F in Bordeaux [southwestern France]; and hospital G in Fort-de-France, French West Indies) reported data for susceptibility to oxacillin and gentamicin of 57,347 S. aureus isolated between 1992 and 1998, after omission of consecutive isolates from the same patients. In addition, hospitals E and G were selected as representative hospitals for exhaustive antibiotic susceptibility analysis of their respective strains. All reported antibiotic susceptibility data were derived from the routine clinical laboratory databases of the participating hospitals and were determined according to the guidelines set by the Committee for Antimicrobial Testing of the French Society for Microbiology (6). Hospitals A to C and E to G used the agar diffusion technique as the antibiotic resistance detection method, whereas hospital D used the ATB expression system (bioMérieux, Marcy l'Etoile, France). Strains were considered resistant to oxacillin if the MIC was >2 μg/ml or if there was a diameter of inhibition of <20 mm around a 5-μg oxacillin disk (6). The disk diffusion assay with oxacillin disks was performed either on Mueller-Hinton agar plates incubated for 24 h at 30°C or on Mueller-Hinton agar plates supplemented with 2% NaCl and incubated for 24 h at 37°C. Susceptibility testing with the ATB expression system was performed according to the instructions of the manufacturer (bioMérieux). Population analysis, in order to determine oxacillin resistance classes of referred isolates, was performed as described by Tomasz et al. (25). Briefly, stationary-phase culture (10⁹ to 10¹⁰ CFU/ml) were plated at seven dilutions (10⁻¹ to 10⁻⁷) on a series of agar plates containing serial twofold dilutions of oxacillin at concentrations ranging from 0 to 1,000 μg/ml. The plates were incubated at 37°C for 48 h before the colonies were counted. The number of bacteria capable of forming colonies was plotted against the concentration of oxacillin, producing population analysis profiles. Statistical analysis was performed with Epi-Info software (version 5; Centers for Disease Control and Prevention, Atlanta, Ga.). One hundred eighty-three MRSA isolates (145 GS-MRSA and 38 GR-MRSA) originating from the seven participating hospitals were randomly selected for molecular typing by pulsed-field gel electrophoresis (PFGE). All strains were kept frozen at −80°C until used.

PFGE.

SmaI macrorestriction patterns were obtained by using a contour-clamped homogeneous electric field system with a CHEF DR-II (Bio-Rad, Richmond, Calif.) apparatus as previously described (9, 13). Comparisons of resolved macrorestriction patterns were based on the recommendations of Tenover et al. (23). Strains differing in up to three fragments only were deemed clonally related and were described as subtypes of a given clonal type. In the case of no differences between banding patterns, strains were considered identical. When they differed by four or more fragments, strains were considered separate types. Letters were used to denote major genotypes, and each variant subtype was indicated by a numerical suffix.

Hybridization studies.

SmaI macrorestriction profiles were vacuum transferred onto nylon membranes (19). Probe labeling and hybridization were carried out by using the nonradioactive digoxigenin DNA labeling and detection kit (Boehringer Mannheim). The GenBank data for the methicillin resistance gene (mecA) (accession no. X52593); the 4′-4"-aminoglycoside nucleotidyltransferase gene (ant4′), responsible for tobramycin resistance (M19465); and the bifunctional 2"-aminoglycoside phosphotransferase–6′-aminoglycoside acetyltransferase gene (aac6′-aph2"), responsible for amikacin, tobramycin, and gentamicin cross-resistance (M18086) were used to design PCR primers allowing the synthesis of gene-specific DNA probes. S. aureus ATCC 6583P was used as a negative control strain for the mecA, ant4′, and aac6′-aph2" genes. S. aureus CIP6525, BM3002, and FK422 were used as positive control strains for the mecA, ant4′, and aac6′-aph2" genes, respectively.

RESULTS

In 1992, the frequency of GS-MRSA within the total population of MRSA was below 7.4% in all of the study hospitals (data were not available for hospital F) (Fig. 1). From 1992 to 1998, this rate progressively increased to reach between 46.8% (hospital D) and 94.4% (hospital G) (Fig. 1). Since the overall incidence of isolation of MRSA remained stable (hospitals A and E), increased slightly (hospitals F and G), or decreased slightly (hospitals B, C, and D) (Fig. 1), this strongly suggested that GS-MRSA strains were supplanting the GR-MRSA population. In the cases of hospitals F and G, the slight increase in the overall incidence of MRSA suggests that the GS-MRSA population arose partly in addition to the endogenous GR-MRSA population.

FIG. 1.

Percentage of S. aureus strains resistant to oxacillin (⧫), and percentages of GS-MRSA (■) and GR-MRSA (□) among MRSA. Mean numbers of S. aureus strains isolated per year: hospital A, 548; hospital B, 1,839; hospital C, 1,321; hospital D, 952; hospital E, 1,188, hospital F, 2,425; hospital G, 612. Data for 1992 and 1993 are not available for hospital F. Data were obtained from a total of 57,347 S. aureus isolates.

For practical reasons, hospitals E and G were selected as a model for studying the evolution of the antimicrobial resistance associated with oxacillin resistance, and the susceptibility data from these hospitals were further analyzed for all MRSA isolates collected in 1996 after omission of consecutive isolates from the same patients. All GS-MRSA strains from hospital E expressed heterogeneous resistance to oxacillin in that, after incubation at 30°C, they showed a clear zone of inhibition around the oxacillin disk and the presence of colonies near the disk, in contrast to all GR-MRSA strains, which showed homogeneous resistance with confluent growth right up to the oxacillin disk (Table 1). Population analysis showed class 1 phenotypic expression for selected GS-MRSA strains in that only a very low proportion (10⁻⁷ to 10⁻⁸) of bacteria could form colonies in the presence of oxacillin at up to 25 μg/ml or more (data not shown) (25). GS-MRSA strains were significantly more frequently susceptible to kanamycin, tobramycin, lincomycin, pristinamycin, tetracycline, minocycline, co-trimoxazole, rifampin, and fusidic acid, and were more frequently resistant to chloramphenicol, than GR-MRSA strains (Table 1). No significant differences between GR- and GS-MRSA in resistance to erythromycin, ofloxacin, and fosfomycin were observed. For hospital G, the analysis of antibiotic susceptibility patterns, while not including exactly the same antibiotics, showed susceptibility profiles similar to those for the isolates from hospital E, except for fusidic acid (Table 1).

TABLE 1.

Antibiotic resistance of MRSA strains (one isolate per patient) isolated in 1996 in hospital E (mainland) and in hospital G (French West Indies)

Antibiotic	Hospital E			Hospital G
	Resistance (%)		P valuea	Resistance (%)		P valuea
	GS-MRSA (158 isolates)	GR-MRSA (102 isolates)		GS-MRSA (122 isolates)	GR-MRSA (75 isolates)
Oxacillin (heterogeneous resistance)	100	0	<0.001	100	0	<0.001
Oxacillin (homogeneous resistance)	0	100	<0.001	0	100	<0.001
Kanamycin	89.6	100	<0.05	77	100	<0.001
Tobramycin	89.6	100	<0.05	77	100	<0.001
Chloramphenicol	52.5	19.2	<0.05	NAb	NA
Tetracycline	8.2	75	<0.001	NA	NA
Minocycline	0	32.7	<0.001	NA	NA
Erythromycin	80.3	86.5	NS	52	92	<0.001
Lincomycin	65.6	84.6	<0.05	22	81	<0.001
Pristinamycin	4.9	19.2	<0.05	1	1	NS
Ofloxacin	91	98.1	NSc	26	97	<0.001
Co-trimoxazole	0	22.2	<0.001	2	1	NS
Rifampin	1.6	26.9	<0.001	1	85	<0.001
Fosfomycin	17.9	11.1	NS	7	24	<0.001
Fusidic acid	1.7	17.6	<0.05	78	14	<0.001

^aCalculated by using Student's t test. 

^bNA, data not available. 

^cNS, not significant. 

Molecular typing analysis using PFGE was performed for 184 randomly selected isolates from the seven participating hospitals. For each isolate, macrorestriction profiles generated by SmaI cleavage comprised 15 to 17 fragments varying in size from <80 to 700 kb (Fig. 2). PFGE profiles of GS-MRSA strains were mostly clustered into three major PFGE types: types A and B, comprising 42% (48 of 114 isolates) and 41% (47 of 114 isolates), respectively, of GS-MRSA strains from the six mainland hospitals, and type C, representing the dominant GS-MRSA type (64% of GS-MRSA strains) in the French West Indies hospital (Table 2). Type A and B strains were seldom detected among isolates derived from the French West Indies, and conversely, type C strains were not detected among isolates derived from mainland hospitals.

FIG. 2.

Representative SmaI-restricted PFGE patterns of MRSA strains. Lane 1, strain A980142 (GS-MRSA, type A.1); lane 2, A960451 (GR-MRSA, type A.7); lane 3, A980157 (GS-MRSA, type A.3); lane 4, A980147 (MRSA strain susceptible to all aminoglycosides, type A.3); lane 5, A960651 (GR-MRSA, type B.1); lane 6, A980400 (GS-MRSA, type B.1); lane 7, A960649 (GR-MRSA, type B.1); lane 8, A980649 (GS-MRSA, type B.1); lane 9, NCTC 8325 (molecular size marker).

TABLE 2.

PFGE types and subtypes of GS- and GR-MRSA) from strains from mainland hospitals A to F and French West Indies hospital G

PFGE		No. of isolates from hospital(s):
		A to D and F, GS-MRSA (n = 55a)	E		G
Type	Subtype(s)		GS-MRSA (n = 59)	GR-MRSA (n = 18)	GS-MRSA (n = 31)	GR-MRSA (n = 21)
A	A.1	5	15	0	0	0
	A.2 to A.8	12	16	7	1	1
B	B.1	21	15	0	0	0
	B.2 to B.7	9	2	0	1	1
C	C.1	0	0	0	12	0
	C.2	0	0	0	7	0
Otherb		8	11	11	10	19

^aHospital A, 8 isolates; B, 16 isolates; C, 14 isolates; D, 5 isolates; F, 12 isolates. 

^bEach type includes fewer than six isolates. In these other types, none of the GS-MRSA types were identical to GR-MRSA types. 

GR-MRSA isolates available from hospitals E and G comprised several types (including type A) and subtypes (Table 2). In several cases, the PFGE profiles of GS- and GR-MRSA strains of the same type differed by two bands only (Table 2 and Fig. 2). Moreover, some other GS-MRSA strains that were resistant to kanamycin and tobramycin had PFGE types strictly identical to those of MRSA strains that were susceptible to all aminoglycosides (Fig. 2). Southern blot hybridizations with the mecA-specific and ant4′-specific probes were positive for the 19 representative isolates of PFGE types A and B, on the same ≈185- to 215-kb SmaI fragment, except for one isolate which was susceptible to all aminoglycosides and showed negative hybridization with the ant4′-specific probe (Table 3). Hybridization with the aac6′-aph2" gene probe occurred on an SmaI fragment of ≈490 kb or above 700 kb for the GR-MRSA isolates only (Table 3), thus confirming the lack of the bifunctional enzyme in GS-MRSA strains.

TABLE 3.

Results of hybridization studies of S. aureus control strains and of selected GS- and GR-MRSA strains

S. aureus strain	Relevant characteristics	Size (kb) of SmaI fragments hybridizing with the following gene probe:
		mecA	ant4′	aac6′-aph2"
ATCC 6583P	mecA ant4′ aac6′-aph2"	NHa	NH	NH
CIP6525	mecA+ ant4′ aac6′-aph2"	≈185–215	NH	NH
BM3002	mecA ant4′+ aac6′-aph2"	NH	≈185–215	NH
FK422	mecA ant4′ aac6′-aph2"+	NH	NH	≈490
A980142	GS-MRSA, type A.1	≈185–215	≈185–215	NH
A960451	GR-MRSA, type A.7	≈185–215	≈185–215	≈490
A980157	GS-MRSA, type A.3	≈185–215	≈185–215	NH
A980147	MRSA strain susceptible to all aminoglycosides, type A.3	≈185–215	NH	NH
A960651	GR-MRSA, type B.1	≈185–215	≈185–215	>700
A980400	GS-MRSA, type B.1, derivative of A960651	≈185–215	≈185–215	NH
A960649	GR-MRSA, type B.1	≈185–215	≈185–215	>700
A980649	GS-MRSA, type B.1, derivative of A960649	≈185–215	≈185–215	NH

^aNH, negative hybridization. 

Spontaneous GS-MRSA derivatives (with no changes in any other antibiotic resistance) were obtained after long-term storage (2 years) at room temperature from two GR-MRSA strains (A960651 and A960649). In contrast, no change from the original GR-MRSA phenotype was observed in strains maintained for the same time period at −80°C. The PFGE types of the parent GR-MRSA strains and revertant GS-MRSA strains were indistinguishable (Fig. 2, lanes 5 to 8). Hybridization studies confirmed the loss of the aac6′-aph2" gene in the GS-MRSA derivatives.

DISCUSSION

S. aureus, and in particular MRSA, has long been one of the more serious and problematic nosocomial pathogens, repeatedly responding to the challenge of new antistaphylococcal antibiotics by acquiring new resistance (3). A more prudent use of antibiotics, in addition to the implementation of better infection control and hygiene measures for reducing nosocomial infections, has been encouraged in many countries and may play a role in reducing the incidence of multiply resistant organisms. Although these implemented measures may have been responsible for the major decline in the incidence of MRSA in Denmark from 15 to 0.2% between 1971 and 1984 (18), they have been relatively unsuccessful in other, larger countries, where the prevalence of MRSA seems to be still increasing (22, 26). It is clear that in most French hospitals, not only hospitals nationwide within the mainland but also a geographically distinct hospital situated in the French West Indies, distinct MRSA clones that are more susceptible to multiple antibiotics, particularly gentamicin, are increasing in incidence, often replacing the endogenous classical MRSA clones, while the overall prevalence of oxacillin resistance among S. aureus isolates remains stable or varies slightly (Fig. 1). This observation confirms the recent reports from two French hospitals (one of which is included in the present study) (2, 12) of the emergence of new epidemic MRSA strains that are more susceptible to antibiotics.

We showed that the same two GS-MRSA PFGE types, types A and B, were predominant in each of the mainland hospitals (A to F) included in the study and that another distinct PFGE type, type C, predominated in hospital G located in the French West Indies. This is probably due to the frequent exchange of patients between hospitals A to F, while exchange between these mainland hospitals and hospital G in the French West Indies is uncommon, explaining the quasiabsence of common mainland MRSA types. Using the definition of a distinct PFGE type as one comprising strains showing no more than a three-fragment variation, a limited number (i.e., 5 to 11) of MRSA types are usually recognized within a given area, always with the predominance of one type. For instance, among 189 MRSA isolates from a Spanish hospital between 1989 and 1993, 83.1% were classified into one among the 11 total PFGE types, whereas 0.5 to 5.3% were classified into the 10 other types (7). Again, 77% of 85 Brazilian MRSA strains from six different hospitals in five cities, isolated between 1992 and 1994, shared minor variants of a common PFGE type (24). Similarly, in Germany the so-called North German epidemic type has been identified and comprises 80% of all isolates in that region (28). The capacity of a dominant clone for geographic spread, at the expense of an already established clone (7, 20), has already been described, but these emergent dominant clones were not susceptible to gentamicin.

At the molecular level we showed that the switch from MRSA with multiple antibiotic resistance to more susceptible MRSA is correlated with the loss of the aac6′-aph2" gene, which confers resistance to all aminoglycosides although it reduces amikacin susceptibility only slightly, with conservation of the ant4′ gene, which confers resistance to kanamycin, tobramycin, and amikacin (27). Since the aac6′-aph2" gene is frequently carried by transposon Tn4001 (4, 8), it is likely that certain GS-MRSA strains could have emerged from GR-MRSA strains by transposon excision or deletion, a phenomenon that we have reproduced in vitro by long-term storage of GR-MRSA strains at room temperature. The fact that both GR- and GS-MRSA strains have closely related PFGE types, which suggests a recent common ancestor for certain GS- and GR-MRSA isolates, supports this hypothesis. It is possible that this phenomenon has resulted from a long-term change in the aminoglycoside selective pressure over a period of 5 years or more (15), as also suggested in the two previous French studies (2, 12). A marked decrease in the use of gentamicin in the Henri Mondor Hospital (located near Paris) between 1983 and 1987 was reported, whereas that of amikacin increased by a factor of two (2); in this hospital GS-MRSA isolates represented 4.9 and 27.5% of the MRSA isolates in 1993 and 1995 respectively. However, data documenting changes in the level of use of aminoglycosides have not been reported for all hospitals. In any case, changes in aminoglycoside use cannot alone explain the increase in susceptibility to other antibiotics (e.g., kanamycin, tobramycin, lincomycin, rifampin, minocycline, and fusidic acid) (Table 1) and the return to heterogeneous resistance to oxacillin observed with the GS-MRSA population. This suggests that some other selective pressure(s) could be involved in the emergence and spread of these strains. The genetic event(s) responsible for this pleiotropic effect on antibiotic resistance is unknown, but the strong association of the loss of the bifunctional enzyme with the return to heterogeneous resistance to oxacillin in strains that likely derive from a recent common ancestor (Fig. 2) suggests that excision of Tn4001 also affected an accessory factor essential for homogeneous-heterogeneous resistance to methicillin. Further studies investigating both the genetic mechanism of and the selective advantage provided by this emerging phenotype should help elucidate the genetic events that are operative.

At a practical clinical level, the continuing emergence of GS-MRSA and the decline in the incidence of GR-MRSA could provide an opportunity for the controlled reintroduction of aminoglycosides for use as anti-MRSA therapies, alleviating the reliance on glycopeptide antibiotics.