Tracking Methicillin-Resistant Staphylococcus aureus Clones during a 5-Year Period (1998 to 2002) in a Spanish Hospital

Abstract

Three hundred seventy-five consecutive methicillin-resistant Staphylococcus aureus (MRSA) clinical isolates recovered between 1998 and 2002 at the Nuestra Señora de Candelaria University Hospital in Tenerife, Spain, were analyzed by molecular fingerprinting techniques to determine MRSA clonal types and their prevalence over time. After determining antibiotic susceptibility, we used SmaI-digested genomic DNA separated by pulsed-field gel electrophoresis (PFGE) to characterize MRSA isolates and to establish PFGE types. Additionally, several selected isolates representative of each major PFGE type were tested by multilocus sequence typing (MLST) and by a multiplex PCR method capable of identifying the structural type of the staphylococcal cassette chromosome mec (SCCmec), generating the corresponding sequence type (ST)-SCCmec types. Results of PFGE, supported by those of MLST and SCCmec typing, allowed us to identify six MRSA clones within the five major PFGE types and document temporal shifts in the prevalence of these MRSA clones from 1998 to 2002. Four of the clones were the pandemic “Iberian” (designated ST247-MRSA-IA), EMRSA-15 (ST22-MRSA-IV), EMRSA-16 (ST36-MRSA-II), and the so-called pediatric (ST5-MRSA-IV) clones, while the other two (ST125-MRSA-IVA and ST146-MRSA-IVA) clones could be derived from the pediatric one. The most striking temporal shift in the dominance of MRSA clones was the replacement of the multidrug-resistant and highly epidemic Iberian clone by the so-called British EMRSA-16 clone during the 5-year surveillance period. Our results are in accordance with previously stated findings showing the worldwide hospital dominance of relatively few pandemic and presumably virulent MRSA clones. We report for the first time the detection in Spain of the British EMRSA-15 and pediatric clones, as well as the abrupt replacement of the Iberian by the EMRSA-16 as the major MRSA clone.

Staphylococcus aureus is the causal agent of most staphylococcal pathologies and is currently a versatile microbial pathogen that has evolved resistance to all antibiotic classes. It is associated with serious community-acquired and nosocomial diseases, although most life-threatening S. aureus infections are hospital acquired (4, 8). Its high level of adaptation to hospital environments has been deeply facilitated by the acquisition of methicillin resistance, an evolutionary step that converted S. aureus to methicillin-resistant S. aureus (MRSA), one of the most common nosocomial pathogens nowadays (19). MRSA emerged with the introduction of an exogenous DNA element into its genome, the staphylococcal cassette chromosome mec (SCCmec), which carries the methicillin resistance mecA gene (16). Recent data shows that acquisition of SCCmec has occurred on multiple occasions and that at least five different methicillin-sensitive S. aureus phylogenetic lineages acquired the element (29). Four main structural types of SCCmec, which differ in size and composition, have been described for S. aureus (14, 15, 20).

Genetic studies using molecular typing methods have shown that most hospital-acquired MRSA infections worldwide are due to any of the so-called epidemic MRSA (EMRSA). These EMRSA clones present great fitness to hospital environments and, consequently, are established in many hospitals and have spread internationally (7). This situation highlights the importance of monitoring the distribution and routes of dissemination of such EMRSA clones at both levels, within hospitals and between distant locations (24). With this purpose, several molecular techniques and an international common nomenclature have been applied to track EMRSA (10, 27). Pulsed-field gel electrophoresis (PFGE) is considered the “gold standard” for establishing clonal relationships at the local level, but its detection capacity seems to make it also too discriminative for global comparisons (5, 31). By contrast, multilocus sequence typing (MLST) has been verified as an adequate method for long-term and global epidemiological studies (11, 48). Combination of MLST with SCCmec typing permits the unambiguous assignment of collections of MRSA isolates to known or new MRSA clones (10).

The aim of this study was to identify MRSA clones circulating in the Nuestra Señora de Candelaria University Hospital (HUNSC), Tenerife, Spain, and to track shifts in their prevalence during a 5-year period (1998 to 2002). For this purpose, we used a combination of different molecular typing methods, including PFGE, MLST, and SCCmec typing.

MATERIALS AND METHODS

Patient features.

Three hundred seventy-five consecutive nonduplicate MRSA clinical isolates were collected from January 1998 to December 2002 from hospitalized patients at HUNSC. The 375 MRSA isolates were obtained from 245 men (65.3%) and 130 women (34.7%). The patients were treated in several wards: 137 (36.5%) in internal medicine, 104 (27.7%) in surgery, 45 (12%) in trauma units, 25 (6.7%) in intensive-care units, 11 (2.9%) in pulmonary, 7 (1.9%) in nephrology, and 46 (12.3%) in other services.

MRSA clinical isolates.

All MRSA isolates recovered in the hospital in each year of the 5-year period were included in the study: 31 isolates (1998), 29 isolates (1999), 57 isolates (2000), 100 isolates (2001), and 158 isolates (2002). The isolates were reported to originate from exudates (174 isolates; 46.4%), nasal cultures (62 isolates; 16.5%), the upper respiratory tract (74 isolates; 19.8%), blood (27 isolates; 7.2%), intravenous catheters (15 isolates; 4%), urine (7 isolates; 1.9%), and other clinical sources (16 isolates; 4.2%). Standard microbiological methods for the identification of S. aureus isolates included Gram staining, colonial morphology, growth on mannitol salt agar, and catalase, coagulase, and DNase testing.

Antimicrobial susceptibility testing.

Antibiotyping was performed with the following automatic systems: from 1998 to 1999, a Vitek 1 system (Card GPS511) was used; from 2000 to 2002, a Vitek 2 system (Card AST-P523) was used (both systems were from bioMérieux, Lyon, France), recording both the reported categories of resistance (susceptible [S], intermediate, and resistant [R]) and the MICs. Seven antimicrobials were chosen to establish the antibiotype of each isolate: penicillin, clindamycin, erythromycin, gentamicin, oxacillin, teicoplanin, and vancomycin. Differences in susceptibility with at least one of these seven antimicrobials characterized a different antibiotype, which was represented by a Roman numeral (Table 1). S. aureus ATCC 25923 was included to check the quality control of the antimicrobial susceptibility patterns. After the initial phenotypic identification as S. aureus and determination of antibiotic susceptibility by microbiological procedures, the isolates were stored in a brain heart infusion broth-glycerol mixture (3:1 [vol/vol]) at −80°C for further molecular analyses.

TABLE 1.

Percentages of MRSA isolates with the different antibiotypes and resistance to clindamycin, erythromycin, and gentamicin

Antibiotype or antibiotic	Susceptibility toa:			% of isolates in yrb:					No. (%) of isolates
	CLIN	ERY	GEN	1998	1999	2000	2001	2002
I	R	R	R	96.8	79.3	49.1	22	19	133 (35.5)
II	R	R	S	3.2	17.2	42.1	73	70.9	215 (57.3)
III	S	R	S	0	3.5	1.8	1	1.3	5 (1.3)
IV	S	S	S	0	0	7	4	8.2	21 (5.6)
V	S	S	R	0	0	0	0	0.6	1 (0.3)
CLIN				100	96.5	91.2	95	89.9	348 (92.8)
ERY				100	100	93	96	91.2	353 (94.1)
GEN				96.8	79.3	49.1	22	19.6	134 (35.7)

^aAbbreviations: CLI, clindamycin; ERY, erythromycin; GEN, gentamicin.

^bIsolates with antibiotypes I to V or resistance to clindamycin, erythromycin, or gentamicin are shown for each year.

Screening for oxacillin resistance.

Phenotypic oxacillin resistance was confirmed by a standard disk method on Mueller-Hinton agar plates with disks each containing 1 μg of oxacillin. Inhibition of growth was interpreted according to National Committee for Clinical Laboratory Standards guidelines (25).

mecA PCR.

In addition to the phenotypic determination of oxacillin resistance, a multiplex PCR assay which permits the simultaneous identification of S. aureus and the detection of the mecA gene was performed as previously described (32, 33). The positive and negative control strains used in mecA detection were S. aureus ATCC 27626 and ATCC 25923, respectively.

DNA agarose blocks preparation and PFGE.

All MRSA isolates were characterized by macrorestriction analysis of SmaI-digested genomic DNA by PFGE. The genomic DNAs were prepared in agarose blocks with the CHEF Bacterial Genomic DNA Plug kit (Bio-Rad Laboratories, Richmond, Calif.) with minor modifications of the manufacturer's instructions.

For restriction endonuclease digestion, each plug was digested overnight with 10 U of SmaI (Promega Corp., Madison, Wis.). Restriction fragments were resolved by PFGE with a CHEF-DRIII contour-clamped homogeneous electric field apparatus (Bio-Rad Laboratories) on a 1% (wt/vol) Seakem Gold agarose (FMC, Rockland, Maine) gel in 0.5× TBE buffer recirculated at 11.3°C. The CHEF-DRIII apparatus was programmed at 200 V (6 V/cm) for 28.5 h, with switching times ramped from 0.5 to 35.0 s. An included angle of 120° was used. Following electrophoresis, the gels were stained with ethidium bromide (0.5 μg/ml), visualized under UV illumination, and photographed with the Gel Doc 2000 system (Bio-Rad Laboratories). Digital images were stored electronically as TIFF files.

Computer-monitored fingerprinting analysis.

Computer analyses of the banding patterns obtained by PFGE were done with the Diversity Database fingerprinting software package, version 2.2 (Bio-Rad Laboratories) after visual inspection. Each gel included two or three lanes with the reference strain S. aureus NCTC 8325 to normalize the PFGE profiles. In general, lanes and bands were automatically assigned by the computer and were corrected manually, after the original images were checked visually by at least two people. A tolerance value of 1% in the band position was applied during comparisons. For cluster analyses, the Dice coefficients were calculated to compute the similarity matrix and transformed into an agglomerative cluster with the unweighted pair group method with arithmetic averages (UPGMA). As in previously published studies, a similarity cutoff of 80% (46) and the criterion of a difference of ≤6 bands (24, 47) were both used to define a PFGE type. Taking into account these criteria, isolates belonging to the same PFGE type were thus designated by one uppercase letter. Within each type, the predominant pattern was considered the first subtype, and isolates with PFGE patterns differing by up to six bands with respect to this subtype were assigned to subsequent subtypes, identified by the corresponding uppercase type letter, followed by Arabic numbers.

MLST and SCCmec typing.

MLST was performed by the methodology described by Enright et al. (11). Sequences were determined with an ABI Prism 310 genetic analyzer with BigDye fluorescent terminator chemistry (Applied Biosystems, Warrington, United Kingdom). Allele numbers were assigned by using the MLST website (http://www.mlst.net). The SCCmec structural types were determined by a multiplex PCR strategy as previously described (30).

Statistical analyses.

Statistical significances of the joint variations of the MRSA clones and the antibiotic resistance along the surveillance period and comparisons between proportions were estimated by bivariant correlations with bilateral significances with the Spearman coefficient. Differences of P < 0.05 were considered statistically significant. Calculations were made with SPSS 11.5 software.

RESULTS

Antimicrobial susceptibility.

The percentage of MRSA increased from 7.8% in 1998 to 16.8% in 2002. Overall, no isolates had reduced susceptibilities to teicoplanin or vancomycin. Hence, the five antibiotypes observed are the results of differences in susceptibility to clindamycin, erythromycin, and gentamicin. Resistance rates to erythromycin and clindamycin remained stable in the period studied. In contrast, the percentage of MRSA isolates resistant to gentamicin decreased drastically, from 96.8 in 1998 to 19.6% in 2002 (P < 0.05) (Table 1).

PFGE clonal types.

Analysis of SmaI macrorestriction profiles of the 375 MRSA clinical isolates revealed 47 unique patterns that clustered above 80% similarity into 21 PFGE types by computer analysis. The same classification was obtained based on number of restriction fragment differences: within each cluster, isolates showed six or fewer band differences from the predominant pattern. The majority of isolates (358 of 375; 95.5%) were clustered in only five major PFGE types (A to E) (Fig. 1; Table 2). Figure 2 presents the PFGE macrorestriction patterns of these five predominant PFGE types. Overall, PFGE type B was found to be predominant (53.4%), followed by PFGE type A (22.4%), PFGE type C (13.3%), PFGE type D (4.8%), and PFGE type E (1.6%) (Table 2). The remaining 16 PFGE types (types F to U) (17 of 375; 4.5%) were termed sporadic and occurred with frequencies of ≤1.5%. In fact, the vast majority of sporadic types (15 of 16; 93.8%) were represented by a single isolate. MRSA isolates belonging to PFGE subtypes A1, B1, and E1 were initially categorized visually by comparison with previously described patterns as the so-called Iberian, EMRSA-16, and EMRSA-15 pandemic clones, respectively (21, 24). These visual determinations were later confirmed by MLST and SCCmec typing as described below.

FIG. 1.

Diversity Database software-generated dendrogram of the 375 MRSA isolates constructed by cluster analysis by unweighted pair group method with arithmetic averages, illustrating similarities based on Dice coefficients of SmaI macrorestriction profiles demonstrated by PFGE. The five major PFGE types (A to E) are indicated on the right. •, sporadic PFGE type J. A discontinuous line indicates the 80% similarity that was used as a cutoff criterion for comparison of PFGE patterns.

TABLE 2.

Genotypic properties of isolates of the five major PFGE types

PFGE type: no. (%) of isolates	PFGE subtype: no. (%) of isolates	MLST profile (no. of isolates tested)	ST	SCCmec type	Previous clone name (ST-SCCmec nomenclature)a
A: 84 (22.4)	A1: 31 (36.9)	3:3:1:12:4:4:16 (2)	247	IA	Iberian (ST247-MRSA-IA)
	A2: 13 (15.5)			IA
	A3: 12 (14.2)			IA
	A4: 6 (7.1)			IA
	A5: 5 (5.9)			IA
	A6: 4 (4.8)			IA
	A7: 3 (3.6)			IA
	A8: 3 (3.6)			IA
	A9: 2 (2.4)			IA
	A10: 2 (2.4)	3:3:1:12:4:4:16 (1)	247	IA
	A11: 1 (1.2)			IA
	A12: 1 (1.2)	3:3:1:12:4:4:16 (1)	247	IA
	A13: 1 (1.2)			IA
B: 200 (53.4)	B1: 172 (86)	2:2:2:2:3:3:2 (2)	36	II	EMRSA-16 (ST36-MRSA-II)
	B2: 18 (9)			II
	B3: 2 (1)			II
	B4: 2 (1)			II
	B5: 2 (1)			II
	B6: 1 (0.5)			II
	B7: 1 (0.5)			II
	B8: 1 (0.5)			II
	B9: 1 (0.5)	2:2:2:2:3:3:2 (1)	36	II
C: 50 (13.3)	C1: 44 (88)	1:4:1:4:12:1:54 (2)	125	IVA	—(ST125-MRSA-IVA)
	C2: 4 (8)	1:4:1:4:12:1:54 (1)	125	IVA
	C3: 2 (4)			IVA
D: 18 (4.8)	D1: 14 (77.8)	1:43:1:4:12:1:10 (2)	146	IVA	—(ST146-MRSA-IVA)
	D2: 1 (5.5)			IVA
	D3: 3 (16.7)	1:4:1:4:12:1:10 (1)	5	IV	Pediatric (ST5-MRSA-IV)b
E: 6 (1.6)	E1: 5 (83.3)	7:6:1:5:8:8:6 (2)	22	IV	EMRSA-15 (ST22-MRSA-IV)
	E2: 1 (16.7)	7:6:1:5:8:8:6 (1)	22	IV

^a—, no previous clone name.

^bIn general, the clone names presented here apply for the entire PFGE type; however, this clone name was applied only to subtype D3.

FIG. 2.

PFGE patterns of SmaI-digested DNA of the five major PFGE types (A to E). The numbers above the lanes correspond to PFGE subtypes (A1 to A13, B1 to B9, C1 to C3, D1 to D3, and E1 to E2). ▿, macrorestriction pattern generated for the S. aureus NCTC 8325 reference strain. Numbers on the right correspond to the molecular sizes (in kilobases) of S. aureus NCTC 8325 SmaI restriction fragments.

Concordance between antimicrobial susceptibility patterns and PFGE types.

Five different antibiotypes were observed. However, the vast majority of MRSA isolates (348 of 375; 92.8%) belonged to antibiotype I or II (Table 1). Except for PFGE types A and B, there was no correlation between antibiotype and PFGE type. There was an excellent correlation between PFGE type A and antibiotype I (97.6%) and between PFGE type B and antibiotype II (87.5%). Nevertheless, the same antibiotype was observed for isolates in more than one PFGE type. Furthermore, variations within the antibiotype associated with a particular PFGE type or subtype were found. We also detected cases in which a given PFGE subtype profile was associated with a particular antibiotype.

As described above, the most drastic change in antibiotic susceptibility was observed with gentamicin (Table 1). Globally, 64.3% (241 of 375) of the MRSA isolates were susceptible to gentamicin, and the majority of these, or 73% (176 of 241), belonged to PFGE type B; 12.5% (30 of 241) belonged to PFGE type C. The remaining 14.5% (35 of 241) were distributed by 16 PFGE types: 0.8% (2 of 241) were PFGE type A, 5.8% (14 of 241) were PFGE type D, 2.5% (6 of 241) were PFGE type E, and 5.4% (13 of 241) belonged to 13 sporadic PFGE types. The progressive decrease of gentamicin resistance from 1998 to 2002 occurred in parallel to an increase in the number of PFGE type B gentamicin-susceptible isolates. In the last period of the study (2001 to 2002), the increase of PFGE type C-susceptible isolates also contributed to the decline of such resistance.

MLST and SCCmec typing.

These two methods were applied for full characterization of several selected MRSA isolates in accordance to the newly proposed nomenclature for MRSA clones (10, 40). Using a recently described multiplex PCR strategy (30), we tested all MRSA isolates for SCCmec types. Overall, SCCmec types IA, II, IV, and IVA were detected in 84 (22.4%), 200 (53.4%), 12 (3.2%), and 68 (18.1%) isolates, respectively. SCCmec types IA and II were identified in the 84 isolates belonging to the PFGE type A and in the 200 isolates belonging to PFGE type B, respectively. SCCmec type IV was identified in the six isolates of PFGE type E, in three isolates of PFGE type D, and in three isolates belonging to sporadic PFGE types. SCCmec type IVA, which differs from type IV by the integration of the linearized pUB110 plasmid, was identified in 47 isolates of PFGE type C, in 14 isolates of PFGE type D, and in 7 isolates of sporadic PFGE types. Although other SCCmec types were not detected, for 11 isolates (2.9%) (3 of PFGE type C, 1 of PFGE type D, and 7 belonging to sporadic PFGE types), SCCmec types could not be inferred by the multiplex PCR strategy, since the amplification patterns observed did not exactly correspond to the ones described by Oliveira and de Lencastre (30), and are being investigated further.

Representative MRSA isolates were also analyzed by MLST (11). Overall, STs were obtained for 17 MRSA isolates. For each major PFGE type, we tested two randomly selected isolates of the predominant subtypes (A1 to E1) and one or two isolates selected as having the greatest number of band differences from the predominant subtypes. The isolates of PFGE types A, B, C, and E tested for MLST showed the STs as ST247, ST36, ST125, and ST22, respectively. Within PFGE type D, two different STs were detected. Isolates of the predominant subtype D1 showed ST146, while the D3 isolate tested, which had a five-band difference from D1, showed ST5 (Table 2). Additionally, we also performed MLST on the single isolate of sporadic PFGE type J, which was the isolate that clustered nearest the PFGE type B, and it displayed ST30.

In summary, we unambiguously ascribe our major PFGE types to ST-SCCmec types (Table 2). The single sporadic isolate tested for MLST was identified as an ST30-MRSA-IV clone. ST30 is a single-locus variant (SLV) and single-nucleotide variant (SNV) of the ST36 characteristic of the EMRSA-16 clone, differing in the pta allele. As previously described, ST247-IA corresponds to the Iberian clone, and ST36-II and ST22-IV correspond to the so-called British EMRSA-16 and EMRSA-15 clones, respectively, confirming our previous presumptive assignations based on SmaI macrorestriction profiles (10). ST5-IV corresponds to the so-called pediatric clone, while ST125 and ST146 are SLVs and SNVs of ST5 differing in the yqiL and aroE alleles, respectively. Remarkably, until now only one isolate belonging to ST125 (strain AA-24236/99, recovered in Norway) and one isolate belonging to ST146 (strain 13ES, recovered in Spain) have been deposited in the MLST database.

Changes in the prevalence of major MRSA clones over time.

Figure 3 illustrates the distribution and evolution over time of MRSA clones observed from 1998 to 2002. In the first year of the surveillance period, the Iberian clone (ST247-MRSA-IA) was found to be the only dominant clone, representing 96.8% of all MRSA isolates. Only one of the 31 isolates (3.2%) recovered that year belonged to another clone, the pediatric clone. In 1999, the Iberian clone was still dominant (79.4% of isolates recovered). Moreover, we detected for the first time in our hospital the British EMRSA-15 clone (ST22-MRSA-IV) (one isolate; 3.4%) and the EMRSA-16 clone (ST36-MRSA-II) (two isolates; 6.9%). In 2000, due to a rapid nosocomial dissemination of EMRSA-16, the Iberian and EMRSA-16 clones were equally represented, at 43.8 and 38.6%, respectively. In 2001 and 2002, there was a drastic decline in the prevalence of the Iberian clone, which was only represented by four (4%) isolates in 2001 and two (1.2%) isolates in 2002. In parallel, from a low level of 6.9% (2 isolates) in 1999, the prevalence of the EMRSA-16 clone increased and in 2001 reached a peak of 74% (74 isolates). However, in 2002 the number of isolates belonging to EMRSA-16 achieved a maximum (102 isolates; 64.6%). Remarkably, in this last period (2001 to 2002), the overwhelming majority of the EMRSA-16 isolates (149 of 176; 84.7%) had the B1 PFGE subtype, representing 57.8% of the total MRSA isolates. Furthermore, we detected the ST125-MRSA-IVA clone for the first time in 2001; representation of this clone increased rapidly from 0% in 2000 to 14% in 2001 and 22.8% in 2002. EMRSA-15 clone prevalence has not increased over time since its detection in 1999, representing only 1.6% (six isolates) during the 5-year surveillance period. In the same manner, PFGE type D, which includes the ST146-MRSA-IVA and ST5-MRSA-IV clones, persisted over time; its percentage of representation during the 5-year period was low and stable (Fig. 3).

FIG. 3.

Temporal shifts in the prevalence of MRSA clones at the HUNSC from 1998 to 2002. n, number of isolates.

DISCUSSION

Epidemiological studies with different molecular typing techniques have indicated that the massive geographic spread of MRSA results from the wide dissemination of a relatively small number of clones (7, 10, 27, 45). Seven major pandemic MRSA (the so-called Iberian, Brazilian, Hungarian, New York/Japan, pediatric, and EMRSA-15 and EMRSA-16 clones) have been identified as causing the majority of hospital-acquired S. aureus infections in the world (29), indicating that they represent successful clones in terms of their ability to cause infection, persist, and spread from one geographic site to another, including across continents. Tracking MRSA dissemination makes it necessary to unambiguously ascribe isolates to previously known clones or to consider them novel clones (10, 40).

The application of PFGE in combination with MLST and SCCmec typing has permitted us to determine the clonal nature of our isolates and document temporal shifts in the prevalence of MRSA clones that occurred in the HUNSC over a 5-year period (from 1998 to 2002). Application of MLST and SCCmec typing in previous studies showed that all major MRSA clones, defined as groups of isolates from more than one country with the same ST-SCCmec type, belong to one of only five clonal complexes (CCs). In this study, the majority of isolates (358 of 375) were assigned to five major PFGE types, including six MRSA clones, distributed into four CCs. Four of these six MRSA clones corresponded to the pandemic Iberian (CC8), EMRSA-15 (CC22), EMRSA-16 (CC30), and pediatric (CC5) clones. The other two MRSA clones could have evolved from the pediatric clone. Moreover, we followed temporal shifts in the prevalence of MRSA clones.

The frequency of MRSA isolated in the hospital changed in a significant way from 7.8% in 1998 to 16.8% in 2002. This rise was accompanied by several interesting events. The replacement of the Iberian MRSA clone by the British EMRSA-16 clone as the major MRSA clone was the most relevant change detected. The multidrug-resistant and highly epidemic Iberian clone (ST247-MRSA-IA) has been one of the most successful contemporary MRSA clones, showing a superior capacity for spreading and replacing other MRSA clones (9, 21, 22, 38, 39, 42, 43, 44). However, there are also studies reporting the replacement of the Iberian clone in hospital settings, for example, by the Brazilian clone (ST239-MRSA-III), which also belongs to CC8 (2, 3, 28). On the other hand, the British EMRSA-16 clone (ST36-MRSA-II) was largely restricted to the United Kingdom, although it has recently been identified in a number of other European countries (6, 24, 34); currently, it is considered a pandemic MRSA clone (29). We detected the Iberian MRSA clone as the only dominant clone in the first surveillance period, 1998 to 1999. Interestingly, our data illustrate a drastic decline in the prevalence of this clone after 2000 and a parallel rise in the representation of the EMRSA-16 clone, showing its high capacity for achieving quick nosocomial spread. As shown in Table 2, PFGE subtype B1 represents the vast majority of EMRSA-16 isolates (86%). Therefore, although we have detected other PFGE subtypes (i.e., B2 to B9), probably originating from subtype B1 during its dissemination, the prominent increase in EMRSA-16 could be mainly the result of a local epidemic due to nosocomial spread of subtype B1 by cross-infection. Thus, this epidemic probably played an important role in the significant increase in the percentage of MRSA infections from 7.8% (1998) to 16.8% (2002). However, the precise mechanism by which the EMRSA-16 clone replaced the Iberian clone after 2000 as the major or dominant clone remains unknown. Since the Iberian clone has been shown to be highly epidemic and virulent, the detection of its replacement by EMRSA-16 is a remarkable finding never described before. Moreover, this phenomenon also involves a change in major lineage, from CC8 to CC30. As described above, there was a correlation between the Iberian clone and antibiotype I and between EMRSA-16 and antibiotype II, reflecting their different susceptibilities to gentamicin. For this reason, the described replacement was associated with a significant decrease in the level of resistance to gentamicin (Table 1). That and other differences in virulence capacity may explain the greater fitness that EMRSA-16 has shown in our hospital. Further studies are being developed to identify the traits that could explain this replacement.

Another important pandemic MRSA clone detected was the British EMRSA-15 clone (ST22-MRSA-IV) (CC22). After emerging in 1991, EMRSA-15 rapidly displaced most of the other MRSA clones in the United Kingdom, where it is now endemic in hundreds of hospitals (18, 26). Currently, EMRSA-15 is one of the two main causes of MRSA infections in the United Kingdom and has also been identified in several other countries (17, 24, 49). In our hospital, the EMRSA-15 clone was detected for the first time in 1999, the same year that EMRSA-16 emerged. Surprisingly, the levels of EMRSA-15 incidence have remained stable, constituting less than 2% of hospital MRSA isolates. To our knowledge, this is the first time that hospital detection of EMRSA-15 clone has been reported in Spain. Further studies are necessary to elucidate why the incidence of EMRSA-15 has remained at a very low level, compared to that of the EMRSA-16 clone, while both clones have shown a high epidemic capacity in hospitals in the United Kingdom and other European countries (17, 23).

Astonishingly, in spite of the great dominance that the EMRSA-16 clone achieved in 2001 after a rapid augmentation began in 1999, the ST125-MRSA-IVA clone (PFGE type C) was detected in 2001 for the first time, and its representation increased rapidly thereafter (Fig. 3). Such an increase, as occurred in the case of EMRSA-16, correlated significantly with the decrease in the level of incidence of the Iberian clone, although ST125-IVA clone levels did not increase enough to mean that the EMRSA-16 did not clearly constitute the major clone. In fact, preliminary analyses of representative 2003 isolates permit us to assume that EMRSA-16 will again be the clearly dominant clone at least in the first year of the next surveillance period (data not shown). Moreover, as also occurred with ST36, our ST125 branch includes gentamicin-resistant and -susceptible isolates, which also has a significant correlation with the decrease in gentamicin resistance, although such a decrease is mainly due to a rise in EMRSA-16-susceptible isolates. Another remarkable aspect is that, although the single ST125 isolate deposited in the MLST database presented SCCmec type IV (Aina Fossum, personal communication), our ST125 isolates were type IVA. Moreover, we detected another clone carrying SCCmec type IVA, ST146-MRSA-IVA (PFGE type D), which persisted at a low, but stable, frequency. Interestingly, within the same PFGE type as ST146-MRSA-IVA, the pandemic pediatric clone (ST5-MRSA-IV) was also identified. ST5 is a very old, globally ubiquitous lineage, predicted to be the ancestor of CC5 (12, 41). As far as we know, this is the first time that ST5 was detected in a hospital in Spain. The interesting detection of ST146 and ST5 within the same PFGE type is not surprising, since similar SmaI restriction patterns have been already observed between isolates with closely related STs (11, 13).

ST125 and ST146 differ from ST5 at the yqiL and aroE loci, respectively. According to information found in the MLST database, ST125 and ST146 are probably derived from ST5, since the single-nucleotide differences in the yqiL allele of ST125 (allele 54) and the aroE allele of ST146 (allele 43) are not found in any other yqiL and aroE alleles, respectively, and therefore must be recent point mutations. Since ST125 and ST146 are SLVs, SNVs of ST5, the ST125-MRSA-IVA and ST146-MRSA-IVA clones described in this study, may have evolved from the pediatric clone (ST5-MRSA-IV) by single-nucleotide substitutions plus a pUB110 integrative event in the mec cassette.

In addition to the 5 major PFGE types, 16 sporadic types were identified. Since they all together only represent 4.5% of all MRSA isolates, we have not considered them relevant to our main objective of tracking the temporal shifts in the prevalence of major clones. Consequently, even though the SCCmec type was obtained for all of them, they were not included in the MLST analysis for this study, although they will be further investigated. However, one sporadic isolate was characterized by MLST; it is the isolate nearest one to the ST36 branch in the PFGE-based tree, showing only seven bands of differences from the B1 subtype. This isolate was identified as a ST30-MRSA-IV clone (CC30). This pattern resemblance is in accordance with the high relatedness of STs from ST30 to ST36, since they constitute SLVs and SNVs, differing in the pta allele. These results enforce the robustness of the cutoff criteria for assigning PFGE patterns. It is remarkable that the isolate representing subtype B9, the one with the largest number of band difference from B1 (six bands), represented ST36, while the sporadic PFGE type J, the nearest sporadic to type B, with a seven-band difference from B1, represented ST30. Hence, our results show an interesting positive correlation between PFGE criteria and genetic relatedness. S. aureus isolates with ST30 are common in the United Kingdom (10). There are 80 ST30 isolates deposited in the MLST database, most of which are methicillin-sensitive S. aureus (83%), with ST30-MRSA isolates in the minority (17%). In several countries (10, 30), a few ST30-MRSA-IV isolates have been found. But there are several reports where ST30-MRSA-IV represents the predominant clone (1, 35, 36, 37). In fact, ST30 is currently considered a particularly successful pathogen, although it does not always carry SCCmec type IV; in particular, community-acquired ST30 presents novel combinations of the SCCmec type (40). In our study, we detected only one isolate belonging to the ST30-MRSA-IV clone, which is susceptible to all non-β-lactam antibiotics, including tobramycin. However, it did not spread in our hospital, where multiresistant MRSA clones are dominant.

In summary, we described clonal diversity and temporal shifts in the prevalence of MRSA clones from 1998 to 2002. Six MRSA clones were identified, four of them being the pandemic Iberian, EMRSA-15, EMRSA-16, and pediatric clones. Moreover, the other two clones probably derived from the pediatric clone, which gives us another reminder of the evolutionary process of the staphylococcal lineages. These results are in accordance with previously stated findings showing the worldwide dominance of a few highly epidemic and presumably virulent MRSA clones. The most striking temporal shift in the dominance of MRSA clones was the replacement of the Iberian by the EMRSA-16 clone as the dominant MRSA clone, which permits us to suppose that EMRSA-16 has had selective advantages under the environmental conditions of the hospital. Furthermore, the other major clones, although persistent or even increased, were not able to avoid the remarked dominance of EMRSA-16 until now. Our study emphasizes the need, recently pointed out elsewhere, for closer international collaboration to monitor the spread of current epidemic strains as well as the emergence of new ones. MRSA surveillance in our hospital is continuing, and it will be epidemiologically relevant to track the evolution of MRSA clones during the following years to better understand the behavior and fitness of this successful pathogen under hospital conditions.