Introduction

Formerly methicillin-resistant Staphylococcus aureus (MRSA) infections were mostly acquired in health care facilities by hospitalized patients [1] and therefore these strains were called hospital-acquired MRSA.

However, in the last years, MRSA also have become an important pathogen in the out-clinic environment [2]. These strains, called community-acquired MRSA (caMRSA), tend to cause infections of the skin and soft tissue and may also be responsible for a severe necrotizing pneumonia [3, 4]. This may be due to the expression of Panton–Valentine leukocidin (PVL, coded by the lukS–lukF-gene), which is able to cause severe necrotic skin lesions in rabbits and lung necrosis in mice [5–8]. PVL-production has also been observed in MRSA associated with hospital infection and in some cases caMRSA without lukS–lukF-gene have been reported [9–11]. Thus, production of PVL is a common feature of caMRSA, but not a defining criterion. However, the Center of Disease Control and Prevention (CDC) describes caMRSA only by anamnestic characteristics (http://www.cdc.gov/mrsa/diagnosis). It has been shown that caMRSA usually harbor SCCmec element type IV and sometimes type V, both smaller SCCmec variants, which carry less resistance information, making caMRSA more susceptible against a number of antimicrobial drugs [6, 12–15]. It has also been shown by several authors that caMRSA strains reported in Europe carry agr-3 in many cases [5, 16, 17], whereas strains presenting agr-1 are found predominantly in the USA [5, 18]. As PVL-producing (PVL+) MRSA are able to cause such severe conditions, it is important to know the epidemiology and prevalence of these bacteria as well as their genetic features. Hence, the aim of the present study was to investigate the prevalence of PVL+-MRSA strains Western Austria and to characterize them.

Materials and methods

Isolates

All Staphylococcus aureus isolates diagnosed in routine laboratory at the Division of Hygiene and Medical Microbiology, Innsbruck Medical University between December 2005 and May 2010 were included in the study. Those, which showed resistance to cefoxitin in agar-diffusion test according to CLSI 2009 guidelines, underwent confirmation using cefoxitin screen-test by the VITEK2 system (bioMérieux, Lyon, France). If positive, these strains were characterized as MRSA. Isolates were stored until further investigation at - 20 °C. A total of 650 MRSA isolates was identified, originating from Innsbruck Medical University hospital (51 %, n = 332), from district hospitals (38.2 %, n = 248), and from general practitioners (10.8 %, n = 70) Western Austria. The MRSA strains were gained from various swabs (skin, nose, throat, ear, and wound), aspirates (joint, ascites), and from blood, sputum, tracheal fluid, and urine. Only one isolate per patient was included in the study.

Presence of the mecA-gene and the lukS–lukF-gene

Presence of the mecA-gene was determined in all investigated strains to verify their classification as MRSA and lukS–lukF was determined to evaluate PVL expression. Both the genes were determined by polymerase chain reaction (PCR) using the GenoType Staphylococcus VER 1.0 kit from Hain Lifesciences (Hain Lifesciences, Nehren, Germany) according to the manufacturer’s instructions [19].

Classification of the SCCmec-element and the agr-gene

SCCmec elements of types I–V were identified by using a combination of different PCRs according to www.staphylococcus.net. The agr-gene of all the lukS–lukF-positive strains was determined using multiplex PCR as described elsewhere [20].

Antimicrobial susceptibility testing

Resistance pattern (clindamycin, cotrimoxazol, erythromycin, fusidic acid, gentamicin, levofloxacin, rifampicin) was evaluated by agar-diffusion testing performed on Müller-Hinton-Agar (BBL; Becton Dickinson, Cockeysville, MD, USA) and using epsilometertest (E-Test®; AB Biodisk, Solna, Sweden) for vancomycin. Plates were incubated for 24 hours at 37 °C. Resistance was interpreted according to the guidelines of the Clinical and Laboratory Standards Institute (Performance Standards for Antimicrobial Susceptibility Testing; Nineteenth Informational Supplement; CLSI, 2009). Concerning fusidic acid, breakpoints published by Olsson-Liljequist et al. [21] were used.

Spa-typing, multilocus-sequence typing (MLST), and arcA-gene

Spa typing was performed according to the Ridom Staph type standard protocol (www.ridom.com) and by the use of the Ridom Staph type software package for assigning spa-types [22].

By the application of the BURP algorithm, spa-types were clustered into different groups, as described by Strommenger et al. [23].

Selected isolates, which were grouped ambiguously by spa-typing, were investigated via multilocus sequence typing (MLST) according to Enright et al. [24] with the exception of an alternative forward primer for the tpi amplimer [25].

All lukS–lukF-positive isolates exhibiting spa-type t008 and sequence type (ST)8 or clonal complex (CC)8 were subjected to PCR for arcA, according to Strommenger et al. [23]. ArcA is indicative for the ACME-element, which has been shown to be present in caMRSA “USA300”.

Results and discussion

Six hundred and fifty MRSA strains (332 from Innsbruck Medical University hospital, 248 from district hospitals, and 70 from general practitioners) were characterized. Presently, this represents the largest PVL+-MRSA survey for Western Austria.

Presence of the mecA-gene and prevalence of the lukS–lukF-gene

All investigated strains carried the mecA-gene (data not shown), confirming their identities as MRSA. Thirty-one of the 650 strains (4.8 %) also harbored the lukS–lukF-gene.

The annual prevalence varied from 0 % in 2005 (only 1 month included in this year) to 8.3 % in 2007. During the study period no marked trend of the prevalence of PVL+-MRSA among MRSA was detectable.

In Innsbruck University hospital, 5.4 % out of 332 isolates (n = 18) were lukS–lukF positive. In district hospitals, 3.2 % out of 248 isolates (n = 8) harbored the lukS–lukF-gene. Twenty-three of the 26 (88.4 %) hospitalized patients were admitted due to (mainly skin and soft tissue) infection with PVL+-MRSA and only one patient acquired the MRSA strain during hospitalization. In the remaining two cases, information about the mode of acquisition of PVL+-MRSA was not available. Among the isolates from general practitioners, 7.1 % out of 70 isolates (n = 5) carried the lukS–lukF-gene (Table 1). The overall prevalence of PVL+-MRSA infections in community-acquired skin and soft-tissue infections with Staphylococcus aureus was 0.3 % in samples admitted to our department during the study period.

Table 1 Characteristics of lukS–lukF-positive MRSA strains
Full size table

Prevalence of the lukS–lukF-positive strains in Europe varies. A Belgian study has investigated 41 MRSA isolates between 2002 and 2004 and has found a lukS–lukF prevalence of 40 % among their isolates. At the first glance, this seems to be extremely high, however, Denis et al. [16] have investigated only MRSA strains originating from community-associated infection. Witte’s research group has found a prevalence of only 0.4 % in Germany in 2003. The reason for this rather low lukS–lukF prevalence may be that the investigated strain collection also contained strains from 221 hospitals [26].

Compared with other data from Austria, we found a rather low prevalence of PVL+-MRSA among our investigated MRSA.

In a survey performed at Vienna General Hospital in 2009, Bauer et al. [27] have found 16 (39 %) out of 41 tested mecA-positive MRSA also harboring lukS–lukF. This rather high prevalence, however, can in part be attributed to a preselection by phenotypic criteria (resistance pattern). Concerning the southeast region of Austria, Grisold et al. [28] have identified 70 caMRSA (13.1 %) among 534 MRSA strains between 2002 and 2007, however, in their survey Grisold et al. observed a tremendous increase of lukS–lukF-positive MRSA from 3.0 (2002) to 24.3 % (2007) during their study period.

Krziwanek et al. [29] have performed their evaluation in five federal provinces of Austria (Upper Austria, Lower Austria, Salzburg, Carinthia, and Vienna). Among 1,150 investigated MRSA strains, they have found 8.2 % (n = 94) positive for lukS–lukF. Also, Krziwanek et al. experienced an increase in their PVL+-MRSA prevalence from 3.7 % in 2005 to 7.7 % in 2006.

Sex distribution in our survey was nearly 50:50: Sixteen (51.6 %) of the lukS–lukF-positive isolates originated from male patients (Table 1).

The average age of patients suffering from PVL+-MRSA was 42.2 years, ranging from 7 years to 93 years. This average age fits well to the data published by others. Jung et al. [30] have reported 42.8 years in their investigation. Estivariz et al. [31] have found a lower median age with only 32 years.

The average age of patients suffering from PVL-negative MRSA in our survey was much higher (62.7 years, ranging from 1 month to 101 years) than in our and the published PVL-positive cohorts [30, 31].

The majority of the lukS–lukF-positive MRSA strains isolated in our laboratory were found in infections of the skin and soft tissue (80.6 %, n = 25). This finding confirms that MRSA strains, which carry the PVL, are likely to cause infections of the skin and soft tissue like already postulated before [32]. Also, the dermatonecrotic potential of PVL is known since a long time [33]. The remaining lukS–lukF-positive isolates originated from blood culture and ear swab (both from patients with infection) and from tracheal fluid and nose swab (both from colonized patients), two samples could not be determined (Table 1).

Classification of the SCCmec-element and the agr-gene

The SCCmec type of all the lukS–lukF-positive isolates was determined. The vast majority (87.1 %) of our tested isolates carried SCCmec IV (n = 11) or variants IVa (n = 14) or IVc (n = 2), one isolate was typed SCCmec I and the SCCmec types of three isolates were not determinable (Table 1). The dominance of SCCmec type IV among our strains fits very well to the data reported by other groups. Denis et al. [16] have investigated the SCCmec of 16 lukS–lukF-positive MRSA strains. Among them they have found all strains carrying SCCmec type IV. Another group has found 16 (72.7 %) out of 22 lukS–lukF-positive MRSA isolates carrying SCCmec IV [17]. Baggett et al. [34] have tested five lukS–lukF-positive MRSA strains and reported that all of them carried SCCmec type IV. These data show that lukS–lukF-positive MRSA strains carry the SCCmec element type IV in most cases, which is also confirmed by our findings.

Investigation of the agr-gene subtype revealed a majority of agr-1 (58.1 %, n = 18). Agr type 3 was found in eight strains (25.8 %). Two strains (6.5 %) were classified agr-2 and one (3.2 %) showed agr-4. In two strains, it was not possible to identify a certain agr-subtype (Table 1).

In contrast to other surveillances in Europe, we have found a very high rate of agr type 1 and only a rather low rate of agr type 3. In Belgium, 13 (81.3 %) of 16 tested lukS–lukF-positive strains have presented agr type 3, only 3 strains (18.7 %) have harbored agr-1 [16]. Naas et al. [17] in France also have found the majority of their tested strains presenting agr-3 (72.7 %, n = 16). Vandenesch et al. [5], who have investigated MRSA strains from three continents, found all lukS–lukF-positive MRSA strains which originated from Europe (France and Switzerland) carrying agr-3, this also holds true for the strains they have received from Australia and Oceania. However, all the lukS–lukF-positive strains in Vandenesch’s study, which have shown agr-1 (n = 4) have originated from the USA, nevertheless, also among the US-derived strains the majority carried agr-3 (87.8 %, n = 29). Tsuji et al. [18] have reported a dominance of agr type 1 among their tested caMRSA strains in the USA.

These differences may be due to the fact that our strains might be a mixture of different genetic backgrounds, likely due to the role of the Tyrol as a touristic hotspot in Western Austria.

Antimicrobial susceptibility

Resistance patterns of all the lukS–lukF-positive strains revealed that none of the tested strains was resistant against vancomycin and rifampicin. Two strains (6.4 %) were resistant against cotrimoxazol (Table 1).

Our data are similar to the data from other research groups who investigated the resistance pattern of the lukS–lukF-positive strains. Also, in Germany [26], in France [17], and in the USA [35], they have found all their strains susceptible against vancomycin and rifampicin. However, they reported their strains also susceptible to cotrimoxazol. Concerning gentamicin and clindamycin, we found three (9.7 %) of our strains resistant, which is higher compared with other data from Europe and the USA. Witte et al. [26] in Germany and Naas et al. [17] in France have reported that all their tested strains were susceptible to gentamicin and clindamycin. Fey et al. [35] in the USA have also found none of their tested strains resistant against gentamicin and a moderate resistance rate (6.3 %) for clindamycin. In our surveillance, four strains (12.9 %) were resistant against levofloxacin, seven (22.6 %) showed intermediate resistance. In another survey, Tsuji et al. [18] found 12 % of their tested caMRSA to be resistant and 88 % sensitive to levofloxacin. Concerning fusidic acid we found 27 strains susceptible and four strains (12.9 %) resistant. This is a sharp contrast to other data from Europe. In Germany and in France, Witte et al. [26 and Naas et al. [17] have described all their investigated isolates as resistant against fusidic acid, however, Witte investigated only ST80 strains. Denis in Belgium has identified 56 % of his isolates as resistant against fusidic acid [16]. In our surveillance, we found 19 strains (61.3 %) resistant against erythromycin and two strains (6.5 %) were detected as intermediate susceptible. In literature, the data dealing with the resistance against erythromycin are quite inconsistent: In the USA, Tsuji et al. [18] have reported 92 % of their investigated caMRSA resistant against erythromycin. Naas in France and Denis in Belgium have found erythromycin resistance rates lower than 20 % among their PVL-positive MRSA: Naas et al. [16]: 19 %, Denis et al. [17]: 6.25 %. Interestingly, Witte et al. [26] found none of their investigated strains resistant against erythromycin. The comparison of our data with the data from other groups shows that there are rather big differences in the resistance rate concerning fusidic acid and erythromycin which may occur because in different nations other antimicrobial drugs might be used and also strains with other genetic background may be predominant.

Spa-typing, multilocus sequence typing (MLST), and arcA-gene

Evaluation of the spa-type and sequence type by MLST method revealed 11 different spa-types (t008, t019, t024, t044, t121, t437, t455, t657, t692, t1201, and t2816) and eight different sequence types or clonal complexes (CC1, CC5, CC8, ST8, CC22, CC30, ST80, and ST88) with ST8 or CC8 (67.7 %, n = 21) and t008 (54.8 %, n = 17) being the most prevalent (Table 1). Thus, there seems to be one large clonal group with 17 isolates containing ten isolates from Innsbruck University Hospital, five isolates from district hospitals, and two isolates from general practitioners. All these strains share the following genetic features: ST8 or CC8, t008, agr-1 (except for two cases where agr could not be determined), and SCCmec IVa (n = 13) or IVc (n = 2; except for two cases, where SCCmec could not be determined). Determination of arcA among t008 strains revealed 13 of 17 (76.5 %) tested strains arcA positive (Table 1). These features give hint to a “USA300” genetic background. USA300 is the most frequent caMRSA strain in the USA but has also spread to Europe [25] and to Asia [36], recently. Furthermore, our population showed two smaller clonal groups consisting of two isolates each: One group shared ST88, t692, agr-3, and SCCmec IV, both isolates originated from Innsbruck University Hospital. The second complex had the following similarities in common: CC30, t019, agr-3, and SCCmec IV. One of these two isolates was found in Innsbruck University Hospital, the other in a district hospital. The remaining ten strains showed no common genetic background and seem to be diverse.

Krziwanek et al. [29] have investigated 94 lukS–lukF-positive MRSA isolates in Austria and have found a majority of them belonging to sequence type ST8 (31.9 %, n = 30). However, they have found 16 strains belonging to sequence type ST152. We found no ST152-strains in our population. On the other hand, Krziwanek et al. [29] have reported 14 strains in their collection belonging to ST80. ST80 also is the second most frequent sequence type in our survey, however with only three isolates. Other data from Belgium showed only three out of 16 strains representing sequence type ST8 and spa-type t008, but eight strains belonging to ST80 and t044 [16]. This shows that also in Europe there is a big variability in the genetic background of the occurring lukS–lukF-positive MRSA strains.

In conclusion, the prevalence of PVL+-strains among MRSA isolates collected in hospitals and general practitioners in Western Austria is 4.8 %. Comparison with other data from Europe is difficult due to unequal criteria defining the study population. Resistance pattern and SCCmec type (mainly IV) was similar to other lukS–lukF-positive MRSA reported from Europe. However, the predominance of agr-1 is special for our population.

The major PVL-positive MRSA lineage found in our population was ST8, t008 which is mainly found in the USA. In contrast, ST80 strains are not found as frequently in Western Austria as in many other European countries.

Acknowledgments

We thank Pfizer Austria for financial support of our study.