Changing Molecular Epidemiology of Methicillin-Resistant Staphylococcus aureus in a Small Geographic Area over an Eight-Year Period

Abstract

The epidemiology of methicillin-resistant Staphylococcus aureus (MRSA) at an international level shows that most MRSA strains belong to a few pandemic clones. At the local level, a predominance of one or two clones was generally reported. However, the situation is evolving and new clones are emerging worldwide, some of them with specific biological characteristics, such as the presence of Panton-Valentine leucocidin (PVL). Understanding these changes at the local and international levels is of great importance. Our objective was to analyze the evolution of MRSA epidemiology at multiple sites on a local level (Western Switzerland) over a period of 8 years. Data were based on MRSA reports from seven sentinel laboratories and infection control programs covering different areas. Pulsed-field gel electrophoresis was used to type MRSA isolates. From 1997 to 2004, a total of 2,256 patients with MRSA were reported. Results showed the presence of four predominant clones (accounting for 86% of patients), which could be related to known international clones (Berlin, New York/Japan, Southern Germany, and Iberian clones). Within the small geographic region, the 8-year follow-up period in the different areas showed spacio-temporal differences in the relative proportions of the four clones. Other international MRSA clones, as well as clones showing genetic characteristics identical to those of community-acquired MRSA (SCCmec type IV and the presence of PVL genes), were also identified but presumably did not disseminate. Despite the worldwide predominance of a few MRSA clones, our data showed that at a local level, the epidemiology of MRSA might be different from one hospital to another. Moreover, MRSA clones were replaced by other emerging clones, suggesting a rapid change.

The burden of Staphylococcus aureus infections is closely related to the bacterium's susceptibly to antibiotics. In 1950, 10 years after the discovery of penicillin, penicillinase-producing S. aureus became a therapeutic problem, especially in hospitals where numerous outbreaks were recorded. In 1959, methicillin, a new betalactam agent resistant to penicillinase, was discovered. Again, soon after the drug's introduction, the first methicillin-resistant Staphylococcus aureus (MRSA) strain was isolated. These strains are resistant not only to methicillin but also to all other antibiotics of the betalactam family, and they are frequently resistant to antibiotics from other classes. Indeed, the mecA gene, responsible for methicillin resistance, is located on a staphylococcal chromosome cassette (SCCmec) that may carry other resistance genes. Five different types of SCCmec (I to V) with several subtypes have been described to date.

Following its emergence, MRSA first caused outbreaks and then became increasingly endemic in many hospitals worldwide. Many studies showed that when a high proportion of MRSA among S. aureus was encountered, one or a few clones were predominant (1, 4, 14, 17, 25, 31, 34, 37, 37, 48, 49, 53). These clones were found not only in a given hospital but also in other hospitals of the same country and even of different countries (2, 2, 5, 33, 44, 47). Analysis of more than 3,000 isolates from Southern Europe, the United States, and South America showed that nearly 70% of them belong to five major pandemic clones, namely the Iberian (sequence type [ST] 247; SCCmec IA), Brazilian (ST 239; SCCmec IIIA), Hungarian (ST 239; SCCmec III), New York/Japan (ST 5; SCCmec II), and Pediatric (ST 5; SCCmec IV) clones (1, 39, 40). At least three more clones should be added to this restricted list to cover Northern European countries: the EMRSA-15 (ST 22; SCCmec IV), EMRSA-16 (ST 36; SCCmec II), and Berlin (ST 45; SCCmec IV) clones (23). It is hypothesized that these clones are particularly transmissible and/or well adapted to the hospital environment (9, 13). New strains with particular virulence were recently found to disseminate in the community (CA-MRSA). They are usually characterized by SCCmec type IV and by the presence of the genes coding for the Panton-Valentine leucocidin (PVL). Genetic analysis of a worldwide collection of these strains showed that they belonged to a few specific clones not encountered in hospitals (21, 50).

However, there have been a few studies reporting the dynamics of MRSA over long periods of time (3, 18, 19, 41, 42, 45, 51). These studies showed that the molecular epidemiology of MRSA is evolving. Whereas in the 1990s, cross-sectional studies showed the predominance of one or two clones in a defined setting, several longitudinal studies have showed replacement of the predominant clones by others within a decade (1, 42).

In 1997, a national surveillance study of MRSA was performed in Switzerland which showed a low prevalence in most hospitals (11). Since then, some clinical microbiology laboratories in Western Switzerland were used to report MRSA patients and isolates on a regular basis to a reference laboratory. The aim of the present report was to study the epidemiology of MRSA clones in different areas of a small geographic region (Western Switzerland) over a long period of time (1997 to 2004). It is of particular interest because it witnesses the increase of MRSA in a region where MRSA rates used to be low (32).

MATERIALS AND METHODS

Setting.

The present report focuses on five cantons of Western Switzerland (Fribourg, Jura, Neuchâtel, Valais, and Vaud), with a total population of 1,422,800 inhabitants (Swiss federal office for statistics, 2004; http://www.bfs.admin.ch/bfs/portal/en/index.html). It comprises 74 hospitals, accounting for a total of 7,664 beds. The University Hospital of Lausanne is an 800-bed hospital and serves as a tertiary care hospital for this region.

Since the national surveillance study in 1997 (11), seven medical laboratories and infection control programs of Western Switzerland report all MRSA patients (infected or colonized) and isolates to a reference laboratory each year. Each laboratory was responsible for the medical microbiological service of several hospitals, other health care institutions, and private practices. These laboratories were considered sentinels. Demographic and epidemiological data (patient's name, age, and sex and date and site of MRSA isolation) were collected.

The 8-year study period was from 1997 to 2004.

Isolates.

All isolates sent to the reference laboratory were reidentified at the species level (catalase test, DNase test, and S. aureus agglutination test [Slidex StaphPlus; bioMérieux, La Balme et Craponne, France]) and checked for methicillin resistance (disk diffusion antibiogram with oxacillin and agglutination test for detection of the PBP2′ protein [Slidex MRSA detection; bioMérieux]) before being stocked at −80°C. In addition, an antibiogram was performed using the Kirby-Bauer method. Based on the inhibition diameters of five antibiotics, antibiotypes were defined as previously described (8, 10). One isolate per patient and per year was analyzed by using molecular typing.

Molecular typing.

Pulsed-field gel electrophoresis (PFGE) was used for molecular typing of MRSA isolates. DNA was digested with the enzyme SmaI, and electrophoresis was performed according to the Harmony protocol (36). Strain NTCC 8436 was used as a migration reference. PFGE gel images were analyzed using the software Bionumerics v. 4.0 (Applied-Maths, Sint-Martens-Latem, Belgium). Two restriction patterns were considered identical when all bands were identical; they were considered related when one to six bands were different (12, 46).

Isolates showing identical or related PFGE patterns were considered to belong to the same clone. Clones were labeled with a capital letter (A, B, C, etc.), and related profiles were indicated by adding a number (A1, A2, B1, B2, etc). PFGE profiles of these strains were compared to a collection of European epidemic strains previously described (6). The type of the SCCmec was determined as previously described (38); selection of isolates was done in order to include for each year the first two to six isolates of each clone.

Further analyses.

At least two isolates per clonal group were analyzed in more detail. Multilocus sequence typing (MLST) was performed as already described (22). STs were assigned according to the existing reference database (http://saureus.mlst.net/). Relatedness to previously described clonal complexes (CCs) was assessed using the eBURST program (http://eburst.mlst.net/).

The accessory gene regulator (arg) allele group (1 to 4) was determined by multiplex PCR as previously described (28). The composition in virulence genes was determined as previously described (28, 29).

RESULTS

General data.

From 1997 to 2004, data for a total of 2,523 patients were reported from all 7 sentinel laboratories and infection control programs of Western Switzerland. Two hundred seven patients were counted twice because they were found to have harbored MRSA on different years. No isolate was available for 474 patients. Of 1,947 patients for whom data were available, 846 (43.5%) were found to have MRSA in screening samples (nose, throat, and inguinal) and 1,101 (56.5%) had clinical samples with MRSA (491 samples from wounds, 299 from urine, 113 from sputa, 39 from blood cultures, and 198 from other sites).

Molecular epidemiology.

Analysis of PFGE data revealed that four predominant genotypes accounted for 86% of all patients. They were called clones A to D. In addition, 4 other minor genotypes were recovered from 20 to 30 patients and were called clones E to H. PFGE profiles of these clones and of some European epidemic clones are shown in Fig. 1. The number of reported patients harboring a predominant clone during the study period is shown in Fig. 2. Additional epidemiologic and genetic characteristics of the 8 clones are shown in Table 1.

FIG. 1.

Similarities of PFGE patterns of clones A to H observed in Western Switzerland between 1997 to 2004 and of some European epidemic clones. ST, CC, and presence of PVL were also indicated when available. Based on the presence or absence of bands, the Dice coefficient was used to calculate the similarities between PFGE patterns.

FIG. 2.

Annual numbers of patients with MRSA reported from the seven sentinel laboratories serving different geographic areas in Western Switzerland. Patients harboring one of the four predominant clones (A, B, C, and D) are also indicated. Since there were large differences in numbers of patients between laboratories, the y-axis scale was adapted accordingly.

TABLE 1.

Epidemiologic and genetic characteristics of the 8 clones isolated from 2,049 patients from Western Switzerland between 1997 and 2004

Clone	ST(s)a	CC	No. (%) of patients from whom clone isolated									% Isolates susceptible to:							Presence of PVLa	Other virulence genesa	agr groupa	SCCmec typeb
			Total	Isolation from screening samples	Isolation from clinical samples						No data	Genta- micin	Cipro- floxacin	Clinda- mycin	Erythro- mycin	Cotrimox- azole	Fucidin	Rifampin
					All samples	Wounds	Urine	Sputa	Blood	Other
A	45	45	329 (16)	92	223	80	90	22	10	21	14	91	17	98	96	99	87	98	−	sec, seil, sem, seio	I	IV
B	105	5	655 (32)	263	379	165	119	33	12	50	13	98	4	36	3	100	94	98	−	sed, seim, seio, seir	II	II
C	247	8	298 (14)	98	165	77	36	13	8	31	35	88	1	31	31	98	97	80	−	sea	I	IVa
C5	8	8	3	1	2	2													+	seik, seiq	I	IV
D	228	5	478 (23)	287	167	74	34	26	8	25	24	2	0	2	2	100	99	92	−	sea, seim, seio	II	I
E	152	Singleton	28	2	24	17	1			6	2	0	74	89	79	100	100	100	+	etv1, edin, hlb	I	New
F	80, new	80	33	7	24	18		3		3	2	97	58	97	71	100	23	100	+	etd, edin	III	IV
G	228	5	28	16	12	4	6	2				0	0	3	3	100	100	97	−	sea, seim, seio	II	I
H	88	Singleton	21	13	8	6				2		100	93	86	71	7	100	100		None	III	IV

^aAnalyses of ST, PVL, other virulence genes, and arg group were based on two to four isolates per clone.

^bFor clone A, 30 tested isolates had SCCmec type IV; for clone B, 23/30 isolates had SCCmec type II, 2/30 type III, and 4/30 type IV; for clone C, 23/45 isolates had type IVA, 11/45 type IV, 10/45 type I, and 1/45 type II; for clone D, 21/21 isolates had type I; for clone E, 7/7 isolates had an untypeable cassette; for clone F, 7/7 isolates had type IV; for clone G, 4/4 isolates had type I; for clone H, 7/8 isolates had type IV and 1/8 was untypeable.

Clone A (ST 45; SCCmec IV).

Clone A accounted for 16% of all patients with MRSA, belonged to ST 45, and harbored SCCmec type IV (CC 45). The predominant PFGE profile accounted for 80% of clone A isolates, whereas related profiles accounted for 20%. This predominant profile was related to the Belgium PFGE group B (16), a Canadian epidemic clone (43), and the Dutch epidemic clone of 2001 (52) and had eight band differences from the Berlin epidemic clone (Fig. 1).

Clone B (ST 105; SCCmec II).

Clone B accounted for 32% of the patients. Its first apparition was in 1998, and a large outbreak (34 patients) occurred in 1999 in the tertiary care hospital. Then, the incidence of patient infections caused by this clone decreased in this hospital (data not shown), possibly as a consequence of a reinforcement of infection control measures. However, this clone again became more frequent between 2001 and 2004. During this period, it was also found in other institutions of Western Switzerland (Fig. 2), and a large outbreak, which occurred in 2001 in several health care institutions, was reported by laboratory no. 2.

The predominant PFGE profile of this clone accounted for 80% of isolates, whereas related profiles accounted for 20%. This clone belonged to ST 105 (Table 1), which was found only once in the international MLST database. ST 105 belonged to CC 5, since it differed from ST 5 by a single base mutation at the yqiL allele. Two isolates of clone B were further typed with the spa typing method (27), and both belonged to spa type t002. The same spa type was observed for isolates of the New York/Japan clone, which also had SCCmec II. This could indicate that clone B derived from this major epidemic clone. It should be noted that 6/30 tested isolates carried different SCCmec types (Table 1).

Clone C (ST 247; SCCmec IV).

Clone C accounted for 14% of the patients and was observed during the entire study period. The predominant PFGE profile accounted for only 30% of isolates, suggesting that this clone did not spread significantly. Whereas the SCCmec types IV and IVa were predominant in this clone, types I and II were also observed (Table 1). Further characterization of two isolates, one from the most frequent PFGE genotype, C, and one from a minority genotype, C5, showed ST 247 (CC8) and ST 8 (CC8), respectively. The subtype C isolate possessed a type IVa SCCmec and was PVL negative (Table 1). The subtype C5 isolate possessed SCCmec IV and was PVL positive. This subtype, C5, was isolated from three patients during the studied period, one in 2001 and two in 2004. The two latter patients had been present in the same emergency room at a less than 3-h interval; nosocomial transmission was highly suspected.

Clone D (ST 228; SCCmec I).

Clone D accounted for 23% of the patients. It was first recovered in 1999, and it was responsible in 2001 for the largest outbreak recorded at the tertiary care hospital (sentinel laboratory no. 1). This clone was also reported at the tertiary care hospital of Geneva (100 km apart from our tertiary care hospital), where it accounted for more than 80% of MRSA bacteremia (J. Schrenzel, personal communication). Surprisingly, in contrast to clones A and B, it was reported more rarely in other laboratories. Two related types, D1 and D2, accounted for 67% and 25% of patients with this clone, respectively. Subtype D1 was predominant at the tertiary care hospital. A third subtype, D3, was predominant in another hospital (laboratory no. 4). Clone D belonged to ST 228 (CC 5) and possessed the SCCmec type I (Table 1), which was previously found in Germany and Slovenia (23).

Clone E (ST 152; SCCmec [new]).

Clone E was recovered from 28 patients in the five cantons of Western Switzerland between 1999 and 2004. Most of these patients came from ex-Yugoslavian countries. This clone belonged to ST 152 and possessed an untypeable SCCmec and PVL. For 61% of these patients, MRSA was isolated from wounds (Table 1).

Clone F (ST 80; SCCmec IV).

Between 1997 and 2004, clone F was recovered from 33 patients of diverse origins (Switzerland, France, ex-Yugoslavia, Rumania, Greece, Palestine, Egypt, Libya, Somalia, and Vietnam). As for clone E, half of the patients had MRSA in a wound (Table 1). This clone belonged to ST 80 (CC 80), and one isolate was found to be of a new ST which had one base difference at allele pta from ST 80 (Table 1). The SCCmec was of type IV, and PVL genes were present. These characteristics were identical to those of the European CA-MRSA clone (50).

Clone G (ST 228; SCCmec I).

Clone G was recovered between 2000 and 2004 from 28 patients originating mainly from Switzerland. It was responsible for two small outbreaks at the tertiary care hospital, each due to a different subtype. The clinical status of these patients showed a majority of carriage and urinary tract infections. Two isolates of this clone showed the same genetic characteristics as clone D (Table 1), but their PFGE restriction profile had 13 band differences from the latter (Fig. 1).

Clone H (ST 88; SCCmec IV).

Clone H was recovered from 21 patients between 2001 and 2004. Most were children from Western Africa coming to Switzerland for surgical operations. These children were screened for MRSA at their arrival in Switzerland, and most of them were already contaminated. All had previous hospitalization histories in their own country. Clone H belonged to ST 88 (Table 1).

Analysis per geographic area served by each sentinel laboratory.

The number of reported patients harboring each of the four predominant clones (A, B, C, and D) from each sentinel laboratory is shown in Fig. 2. Relative frequencies of these clones changed over time in the different areas served by the laboratories. At the tertiary care hospital (sentinel laboratory no. 1), clone A had been predominant since 1996 (9) and was progressively replaced by clones B in 1999 and D in 2000. The increase of MRSA patients at the tertiary care hospital was linked to the appearance of new clones and their dissemination. Data from the other sentinel laboratories (no. 2 and 3) of the canton of Vaud showed that clone A was replaced by clone B and that clone D represented only a minority of patients. In Valais (sentinel laboratory no. 6), clone A was predominant in 1997, and its relative frequency decreased until 2000, the last year for which typing data were available. In Neuchâtel (sentinel laboratory no. 7), according to available data, the frequencies of clones were also changing: clone C was partially replaced by clone B. In the two last areas where the number of MRSA patients were low (Fribourg, sentinel laboratory no. 4, and Jura, sentinel laboratory no. 5), a high diversity of genotypes was found, and despite the presence of the four predominant clones, no trend towards clone replacement could be observed.

DISCUSSION

Our study showed that even in a small geographic region of about 1.4 million inhabitants, the epidemiology of MRSA was different from one area to another. The relative frequency of clones was found to be different from one area to another and changed over time. Within the same period of time, in the same setting, clones B and D were rising whereas clone A was fading. In addition, we witnessed the apparition of new clones (E, F, G, and H), some of them with enhanced virulence (E and F), that did not appear to spread during the studied period. Thus, it is probably not possible to extrapolate the MRSA epidemiology from one site to another or from one period of time to another. Although our data do not demonstrate changes in MRSA rates, the increase in the relative frequencies of new clones (B and D) over time was correlated with an increase in the number of reported MRSA patients and in the proportion of MRSA among S. aureus strains (data not shown).

CA-MRSA.

CA-MRSA represents a new challenge in terms of pathogenicity and antibiotic susceptibility. These strains are usually spreading in the community and are genetically different from hospital-acquired MRSA. The majority of CA-MRSA strains are characterized by the presence of PVL and SCCmec type IV. Complementary genetic analyses showed that three of the clones encountered in our region shared the same genetic characteristic as CA-MRSA (clones C5, E, and F). However, we did not have data on hospitalization histories of patients harboring these clones to confirm the origin of MRSA.

Data on the origin of patients harboring clone E (ST 152) strongly suggest that this clone was associated with immigrants from the ex-Yugoslavian countries. ST 152, PVL positive, was previously reported in only 1/12 Slovenian football team members who suffered from severe soft-tissue infection (35). The second clone (F) was identical to the European CA-MRSA clone (ST 80; SCCmec IV) (20). The last putative CA-MRSA strain was of the subtype C5 (found in three patients). PFGE and MLST data suggest that the genetic background of subtype C5 was similar to that of the Iberian pandemic clone. However, with the presence of the PVL genes and SCCmec type IV, these strains shared genetic characteristic with CA-MRSA. Similar CA-MRSA strains (ST 8; SCCmec IV) have already been reported in Belgium (15). This highlights the fact that PVL-positive MRSA could also have emerged from a genetic background similar to those encountered in the hospital setting.

International dissemination of MRSA and pandemic clones.

The presence of clone E in immigrants highlights a nonnegligible cause of worldwide dissemination of MRSA. The second example of dissemination was illustrated in our study by patients harboring clone H, since the majority of them were children from Western Africa transferred to Switzerland for surgical operation. The importance of immigration of such MRSA clones should nevertheless be put in perspective. Of the three emerging clones E, F, and H, none was responsible for an outbreak in our region. In addition, similar to the case with clone C, PFGE subtypes showed a greater diversity than those for clone A, B, or D, suggesting that importation of these strains was more important than dissemination in our region. From these data, we estimated that these new clones (including CA-MRSA) were not yet a burden in our region. Migration of people should nevertheless be considered in the prevention and control of MRSA, since it might play an important role in the introduction of new clones. It was possibly this way that clone D, which was previously encountered in South Germany and Slovenia, was introduced in our region.

With regard to pandemic clones, if one considers as a clone all isolates belonging to the same ST and having the same SCCmec type (a less-stringent definition than the one we used in our study), two of the epidemic clones in our region belong to previously described pandemic clones. Clone A belonged to the Berlin clone and clone B to the New York/Japan clone. Furthermore, clone D probably belonged to another known epidemic clone, the Southern Germany clone.

Surprisingly, we observed other pandemic clones which did not widely disseminate in our region. The PFGE profiles of clone C isolates were identical or related to those of isolates from the Iberian clone from Portugal, Belgium, and Spain (Fig. 1), which is well known to be highly epidemic and multiresistant in many European countries. In our region, clone C did not seem to disseminate, since the analysis of the diversity of subtypes of clone C (one to six band differences) showed no predominant PFGE profile. It suggested to us that isolates from this clone were constantly imported in our region. However, it is still not clear if clone C should be considered similar to the Iberian clone, since the latter possessed the SCCmec type I or Ia, while our clone C had the type IV and IVa cassette. Other examples of nondissemination of known pandemic clones were EMRSA-15 and -16, which were isolated only on five and three occasions, respectively (data not shown).

Epidemiological evolution of clones.

Besides our observation of the emergence and fading of a clone (clone A) in our region, there are several similar reports. The first report was probably the replacement in England of EMRSA-1 by EMRSA-15 and -16 (13). Recent reports showed that on two occasions, the Iberian clone was replaced by another pandemic clone. It was replaced by EMRSA-16 in one Spanish hospital, while the rate of MRSA among S. aureus strains remained constant (42), and by the Brazilian clone in one Portuguese hospital (3). The fact that on both occasions the Iberian clone was replaced might suggest that it had lost its epidemic potential during the last decade. In a Mexico City hospital, a local clone (ST 30; SCCmec IV) was completely replaced over a 2-year period by the New York/Japan clone (ST 5; SCCmec II) (51). Changes in clones might have significant medical consequences, since the new clones often display different antibiotic susceptibility patterns (3, 7, 16, 41, 42, 45, 51). If we hypothesize the presence of fitness islands in S. aureus (26), the waxing and waning of clones could be explained by recombination events of these islands.

The question raised by these observations is whether or not strains from epidemic clones that were isolated in different countries and at different periods of time have the same biological fitness characteristics. For instance, are the strains of EMRSA-16 isolated in Western Switzerland as fit as EMRSA-16 strains isolated in England in the 1990s?

These reports raise the question of clone definition. Whereas the evolutionary significance of CCs defined by MLST data for S. aureus has been established (22, 24, 30), the biological or epidemiological relevance of clones has never been investigated. Grouping of isolates in clones depends on the markers used (e.g., PFGE, spa typing, and MLST). For instance, the Mexican clone differed by only one band in its PFGE profile from the EMRSA-16 clone. Whereas the former belonged to ST 30 and had the SCCmec cassette type IV, the latter belonged to ST 36 and had the SCCmec type II (51). On the other hand, clone G, described in this study, showed genetic characteristics similar to those of clone D (same ST, same SCCmec type, and same virulence gene content) (Table 1) but had 16 band differences in the PFGE restriction profile. To us, it suggested that this grouping had neither biological nor epidemiological relevance.

MLST provides probably the most robust subtyping system for S. aureus. MRSA clones can be defined as all isolates belonging to the same ST. However, this might not be discriminatory enough to differentiate all recognized clones. For instance, ST 247 includes both the Brazilian and the Hungarian clones, and ST 5 includes the New York/Japan and Pediatric clones. These clones were differentiated by other typing methods, such as PFGE and SCCmec typing (1). Obviously, further research is needed to describe the internal structure of clonal complexes and the evolution of elements such as the SCCmec in order to have a meaningful definition of clones. Such studies will have to rely on knowledge of well-documented local epidemiology of MRSA, based on highly discriminatory and definitive typing methods.

Conclusions.

The epidemiology of MRSA is evolving rapidly. New clones emerge and replace previously predominant clones. These clone dynamics were not uniform, however, and we have shown that even on a regional scale, the MRSA epidemiology was evolving differently from one site to another. Continuous surveillance of MRSA, with the help of a molecular typing method, is of great importance for understanding the local epidemiology of MRSA.