CC8 MRSA Strains Harboring SCCmec Type IVc are Predominant in Colombian Hospitals

Background Recent reports highlight the incursion of community-associated MRSA within healthcare settings. However, knowledge of this phenomenon remains limited in Latin America. The aim of this study was to evaluate the molecular epidemiology of MRSA in three tertiary-care hospitals in Medellín, Colombia. Methods An observational cross-sectional study was conducted from 2008–2010. MRSA infections were classified as either community-associated (CA-MRSA) or healthcare-associated (HA-MRSA), with HA-MRSA further classified as hospital-onset (HAHO-MRSA) or community-onset (HACO-MRSA) according to standard epidemiological definitions established by the U.S. Centers for Disease Control and Prevention (CDC). Genotypic analysis included SCCmec typing, spa typing, PFGE and MLST. Results Out of 538 total MRSA isolates, 68 (12.6%) were defined as CA-MRSA, 243 (45.2%) as HACO-MRSA and 227 (42.2%) as HAHO-MRSA. The majority harbored SCCmec type IVc (306, 58.7%), followed by SCCmec type I (174, 33.4%). The prevalence of type IVc among CA-, HACO- and HAHO-MRSA isolates was 92.4%, 65.1% and 43.6%, respectively. From 2008 to 2010, the prevalence of type IVc-bearing strains increased significantly, from 50.0% to 68.2% (p = 0.004). Strains harboring SCCmec IVc were mainly associated with spa types t1610, t008 and t024 (MLST clonal complex 8), while PFGE confirmed that the t008 and t1610 strains were closely related to the USA300-0114 CA-MRSA clone. Notably, strains belonging to these three spa types exhibited high levels of tetracycline resistance (45.9%). Conclusion CC8 MRSA strains harboring SCCmec type IVc are becoming predominant in Medellín hospitals, displacing previously reported CC5 HA-MRSA clones. Based on shared characteristics including SCCmec IVc, absence of the ACME element and tetracycline resistance, the USA300-related isolates in this study are most likely related to USA300-LV, the recently-described ‘Latin American variant’ of USA300.


Introduction
Since its emergence in 1961, methicillin-resistant Staphylococcus aureus (MRSA) has traditionally been considered a nosocomial pathogen. In recent years, however, community-associated MRSA (CA-MRSA) has emerged as an important global public health problem [1]. Healthcare-associated MRSA (HA-MRSA) infections generally occur in individuals with predisposing risk factors such as surgery or presence of indwelling medical devices, whereas CA-MRSA infections typically occur in otherwise healthy individuals who do not exhibit such risk factors [2]. Eleven SCCmec types (SCCmec I-XI) have been described to date [3,4]; of these, SCCmec types I, II and III are characteristic of traditional HA-MRSA strains, while types IV, V and VI are generally associated with CA-MRSA [1].
In Colombia, MRSA constitutes an increasingly worrisome clinical problem [5,12]; however, understanding of its molecular epidemiology remains limited. Several HA-MRSA clones have been described, including the Pediatric clone (ST5-MRSA-IV, currently classified as SCCmec type VI [16]), the Brazilian clone (ST239-MRSA-III) and the Chilean clone (ST5-MRSA-I), mainly in large tertiary hospitals in Bogotá, the capital city [17,18]. In addition, the CA-MRSA clone USA300 (ST8-MRSA-IV) has recently been associated with nosocomial infections in Latin America [5,12,19]. Most of the latter isolates belong to a distinct Latin American variant of USA300 recently dubbed ''USA300-LV'', which is characterized by carriage of SCCmec IVc, absence of the arginine catabolic mobile element (ACME) and high prevalence of tetracycline resistance [12,20]. Considering the dynamic nature of MRSA epidemiology, it is imperative to evaluate the situation in other regions of Colombia where MRSA constitutes a severe problem. Accordingly, the aim of this study was to evaluate the molecular epidemiology of MRSA strains isolated from patients between 2008-2010 in three tertiary-care hospitals in Medellín, the second largest city in Colombia.

Study Population
An observational cross-sectional study was conducted at three tertiary-care hospitals from February 2008 to June 2010. Hospital A is a large 648-bed university hospital; Hospital B is a 380-bed medium-size tertiary care center; and Hospital C is a 140-bed cardiology hospital. All institutions are located in Medellín, the second largest city in Colombia.

Clinical and Epidemiological Data
Clinical and epidemiological information was obtained from medical records for each patient. Infections were classified as either CA-or HA-MRSA, according to standard epidemiological definitions established by the U.S. Centers for Disease Control and Prevention (CDC) [21]. Healthcare-associated infections were further classified as either community-onset (HACO) or hospitalonset (HAHO) [21]. Infections were classified as HACO if (i) positive MRSA culture was obtained within the first 48 hours of hospital admission and (ii) at least one of the following healthcareassociated risk factors was present: presence of an invasive device at time of admission, history of MRSA infection, surgery, hospitalization, dialysis, or intensive care unit (ICU) admission during the 12 months preceding the culture date. Infections were classified as HAHO if (i) positive MRSA culture was obtained 48 hours after hospital admission and (ii) at least one of the abovementioned risk factors was present. Lastly, infections were defined as community-associated (CA) if positive MRSA culture was obtained during the first 48 hours of hospital admission without healthcare-associated risk factors.
PCR confirmation of S. aureus and methicillin resistance. Presence of the species-specific nuc and femA genes, as well as the mecA gene (determinant of methicillin resistance), were verified by polymerase chain reaction (PCR) as previously described [23,24].
Spa typing and multilocus sequence typing (MLST). The polymorphic X region of the protein A gene (spa) was amplified and sequenced as previously described [27]. Corresponding spa types were assigned using eGenomics software, as described previously [27,28]. Ridom spa-types were subsequently assigned using the spa typing website (http://www.spaserver.ridom.de/) developed by Ridom GmbH and curated by SeqNet.org (http://www.SeqNet.org/) [29]. MLST was performed on a representative subset of 54 isolates (,10% of all samples) using the methodology described by Enright et al. [30]. Allele numbers and sequence types (ST)s were assigned using the database maintained at http://saureus.mlst.net/, while clonal complexes (CC)s were inferred using eBURST analysis [31]. Clonal complexes for all remaining strains were inferred by spa repeat pattern analysis [28], or by referring to the Ridom SpaServer website.
Pulsed-field gel electrophoresis (PFGE). PFGE following SmaI digestion was performed according to a protocol described elsewhere [32]. DNA fragment patterns were normalized using S. aureus strain NCTC8325. Band assignments were manually adjusted after automatic band detection and only bands ranging from 36 kb to 600 kb were utilized for analysis. Cluster analysis was performed using the Dice coefficient in BioNumerics software version 6.0 (Applied Maths, Sint-Martens-Latem, Belgium). Dendrograms were generated by the unweighted pair group method using average linkages (UPGMA), with 1% tolerance and 0.5% optimization settings. Similarity cutoffs of 80% and 95% were used to define types and subtypes, respectively [32]. Representatives of the most common MRSA clones described in Colombia (Chilean, Pediatric and USA300) were used as reference strains.
Statistical analyses. Categorical variables were compared using Chi-square or Fisher's exact tests, with p values #0.05 considered statistically significant. Statistical analyses were carried out using the software package SPSSH v15.0 (SPSS Inc., Chicago, USA).

Clinical and Epidemiological Data
A total of 538 MRSA isolates were obtained from patients admitted to all three hospitals, including 345 patients from Hospital A, 156 patients from Hospital B and 37 patients from Hospital C. Among the patients included in the study, 355 (66.0%) were males, while the median age among all patients was 39 years (range 0 to 92 years). Using CDC criteria, 243 (45.2%) isolates were classified as HACO-MRSA, 227 (42.2%) as HAHO-MRSA and 68 (12.6%) as CA-MRSA. Variations in the prevalence of the above classifications were found in each institution evaluated (Table 1). Hospitals A and C had the highest prevalence of HAHO-MRSA (48.1% and 51.4% respectively), while Hospital B had the highest prevalence of HACO-MRSA (59.6%).
Twenty isolates from each of the most predominant CC8 spa types were further analyzed by PFGE (Figure 3). Isolates belonging to spa type t1610 and t008 were closely related to the USA300-0114 CA-MRSA clone (ST8-MRSA-IVa), with similarity coefficients ranging from 80-85%, whereas the t024 isolates did not appear to be related to this strain (similarity coefficient 73%). However, when PFGE was performed simultaneously on all three spa types, they all appeared to be closely related to USA300-0114 (.80% similarity coefficient) ( Figure 4). Moreover, they did not cluster separately by spa type when analyzed together. As expected, the spa type t149 (CC5) isolates were closely related (similarity coefficients ranging from 85-97%) to the Cordobes/ Chilean clone (ST5-MRSA-I), whereas isolates related to the previously reported Pediatric clone (ST5-MRSA-IV, now classified as SCCmec type VI) were not found (data not shown).
The isolates displayed distinct virulence gene patterns according to clonal complex and SCCmec type, with CC8 isolates harboring SCCmec types IVa, IVc and V more likely to possess genes coding for virulence factors. The genes for PVL (lukS/F-PV) were present in isolates harboring most SCCmec types (except types II and IVg) and were most frequently observed in isolates harboring SCCmec type IVc (90.6%). Exfoliative toxin gene A (eta) was only detected in a single CC8 strain harboring SCCmec type V, while exfoliative toxin gene B (etb) was not detected at all. Various staphylococcal enterotoxin genes were also associated with different SCCmec types, with seb and sed being the most common. The ACME-arcA gene, frequently associated with USA300-0114 strains, was not detected in any of the isolates tested.

Discussion
Molecular epidemiology studies have highlighted the continuing global evolution and spread of different MRSA clones [35]. The factors contributing to dissemination of MRSA clones are only partially understood, but are thought to include migration of human populations, ineffective methods to control MRSA transmission and inadequate treatment strategies, including the inappropriate selection and use of antibiotics [36]. Since 2003, many reports have documented the spread of SCCmec type IVharboring CA-MRSA strains in hospital settings, primarily in Europe and the United States (U.S.), but also in South America [5,6,7,8,9,10,12,13,14,15]. Partly as a result of this incursion, the epidemiologic distinction between CA-MRSA and HA-MRSA has become increasingly blurred. Understanding of this phenomenon remains limited in Latin America, mostly due to a relative lack of studies documenting the current epidemiology of MRSA in institutions within these countries. Accordingly, this study focuses on the epidemiology of MRSA in three important tertiary-care hospitals in Medellín, Colombia.
Our findings confirm a high prevalence of SCCmec type IVcbearing isolates (traditionally associated with CA-MRSA) in local hospitals, where they are currently circulating as nosocomial pathogens along with HA-MRSA strains harboring SCCmec type I.
In this study, SCCmec type IVc was predominant in isolates classified as CA-MRSA (92.4%) and HACO-MRSA (65.1%), while among HAHO-MRSA the frequency of SCCmec type IVc (43.6%) was nearly as high as that of SCCmec type I (48.2%). Moreover, differences in the prevalence of SCCmec type IVc were found among the three hospitals in our study. For example, in Hospital B, SCCmec IVc was found more frequently among HAHO-MRSA isolates (72.5%), whereas in Hospital C, HA-MRSA isolates (both HACO and HAHO) mainly harbored SCCmec type I (47.1% and 54.7%, respectively). In our study, the emergence of SCCmec type IVc in hospital settings was primarily observed in CC8 strains exhibiting three spa types (t1610, t008, t024). Data collected over the course of the study highlight the expansion of these strains, which increased in prevalence from 2008 (46.5%) to 2010 (64.3%). . Genetic relatedness between CC8 MRSA isolates, stratified by spa type. UPGMA dendrogram showing genetic relatedness between representative CC8 MRSA isolates in this study, as determined by PFGE with SmaI and stratified according to spa type: (A) spa type t1610 isolates, (B) spa type t008 isolates and (C) spa type t024 isolates. The broken line corresponds to the cutoff level (80%) used to define related PFGE clones. doi:10.1371/journal.pone.0038576.g003 SCCmec type IVc has been found in other widely disseminated CA-MRSA clones, such as the Oceania-South West Pacific clone (ST30-MRSA-IVc), which was initially described in New Zealand [37] and subsequently isolated from nearly every continent [2]. In South America, it has been reported in CC30 strains from Brazil, Uruguay and, more recently, Argentina [6,13,38,39,40,41], while several reports have described an SCCmec type IVc-bearing variant of USA300 in Colombia, Ecuador, Venezuela, Peru, Argentina and Trinidad and Tobago [5,20]. This Latin American variant, recently dubbed ''USA300-LV'' [20], is distinguished from prototypical USA300-0114 strains in that it harbors SCCmec IVc rather than IVa, lacks the ACME element and is frequently resistant to tetracycline. Although USA300-LV control strains [12] were not available for comparative analysis, the SCCmec IVcharboring USA300-related strains in our study also possessed these characteristics and are most likely representatives of the USA300-LV lineage.
Currently, t008 (CC8) is the fourth most common spa type in the Ridom database (5.93%) and is represented in most European countries, Argentina, Canada and especially the U.S., where it accounts for the majority of USA300 CA-MRSA isolates [1]. It is also the primary spa type associated with USA300-LV, which has been reported in several countries outside of Latin America, including Spain, Italy, Belgium, the Netherlands, the United Kingdom and Australia [20]. Spa type t024 (CC8), while less common than t008 (0.78% in Ridom), is also widespread globally, particularly in Denmark [1]. By contrast, in our study the most common spa type (25.7%) among all 538 MRSA isolates was t1610 (CC8), which has been isolated in Spain, Germany, Sweden, Norway, Canada and the U.S., but has a low overall frequency in the Ridom database (0.01%). Lastly, t149 (CC5), associated primarily with the SCCmec type I isolates in our study, is one of the most common spa types in South America, including Argentina, Brazil, Chile and Paraguay and is related to the previously described Cordobes/Chilean clone [42,43].
Overall, the single most prevalent lineage in this study was ST8-MRSA-IVc (55.0%), followed by ST5-MRSA-I (31.7%). By contrast, an earlier report from Colombia [18] found that MRSA strains isolated from 1996-1998 at five hospitals in Bogotá displayed 80% homology to the Pediatric clone (ST5-MRSA-IV, currently classified as VI), which is characterized by a multiresistant phenotype. However, this lineage was completely absent from the isolates analyzed in this study. In a subsequent study conducted by Cruz et al. [17], the Cordobes/Chilean clone (ST5-MRSA-I) was shown to be the dominant nosocomial strain from 2001-2003. More recently, an ST8-MRSA-IV clone related to USA300 (i.e., USA300-LV) was described as causing CA-MRSA infections in Colombia [44], with reports of nosocomial infections surfacing in 2009 and 2010 [5,12]. In this study, the prevalence of the ST8-MRSA-IVc clone appeared to increase over time, such that presently it is the predominant clone causing both CA-MRSA as well as HA-MRSA infections.
The most prevalent lineages observed in our study were notably different from those found in other South American regions, where ST5-MRSA-II, ST5-MRSA-IV, ST30-MRSA-IV and ST239-MRSA-III appear to predominate [19]. In addition, the Cordobes/Chilean ST5-MRSA-I clone, highly prevalent in South American HA-MRSA infections, is being displaced by the ST8-MRSA-IVc clone in Medellín. These results highlight the geographic diversity among MRSA strains and the predominantly Figure 4. Genetic relatedness between CC8 MRSA isolates, without stratification. UPGMA dendrogram showing genetic relatedness between representative CC8 MRSA isolates in this study as determined by PFGE with SmaI, without stratification by spa type. Note that t008, t024 and t1610 isolates do not cluster separately by PFGE. doi:10.1371/journal.pone.0038576.g004 clonal evolution of healthcare-associated MRSA in Colombia, with successive clonal replacement observed over time. Moreover, they emphasize the importance of local surveillance and dissemination of findings, in order to ensure that medical personnel are aware of changing patterns of MRSA epidemiology within their institutions, thereby enabling them to choose efficacious empirical treatments. In addition, these results, along with similar reports from hospitals in various different countries [5,32,33,45,46], suggest that the use of SCCmec type IV as a genetic marker for CA-MRSA is of increasingly limited significance.
The differences observed in spa typing and virulence gene content, in conjunction with the results of PFGE, suggest the presence of considerable genetic diversity among the ST8-MRSA-IVc strains that have become established in Medellín. In this study, we performed spa typing prior to PFGE, then compared different spa-based PFGE clusters to USA300-0114, as well as to other clones previously reported in Colombia. As shown in Figure 3, whereas the t1610 and t008 isolates were closely related to USA300-0114 (.80% similarity), the t024 isolates did not appear to be related (similarity coefficient 73%). However, when analyzed together (Figure 4), all three spa types were closely related to USA300-0114 by PFGE (.80% similarity) and did not cluster separately. These results are similar to those of a previous study from Denmark [45], in which t008 and t024 isolates were indistinguishable by PFGE. In this study, the t024 strains were genetically distinct from USA300-0114 by PFGE, but possessed the same characteristics as t1610 and t008, including SCCmec IVc, absence of ACME and tetracycline resistance. It is therefore likely that all three ST8-MRSA-IVc spa types belong to the USA300-LV lineage. Nevertheless, in accordance with Larsen et al's recommendations [45], spa typing allowed for differentiation of clones which could not be discriminated using PFGE alone. These findings are relevant since studies performed in Colombia and South America typically report the dissemination of USA300 based solely on PFGE [5,12], whereas our results also suggest that additional methods such as spa typing should be performed to differentiate genetically and epidemiologically distinct strains.
Taken together, the data from this study suggest that CC8 MRSA strains harboring SCCmec IVc are becoming predominant in Medellín hospitals. In addition to the previously described spa type t008 USA300 strains [5,12], two novel ST8-MRSA-IVc spa types (t024 and t1610) have been identified for the first time in Latin America. Given the transmissibility and potential virulence of CA-MRSA strains, continued increase in the prevalence, severity and complexity of infections caused by these emerging clones may lead to greater morbidity, mortality and hospitalization costs [11,47]. Rigorous surveillance is therefore necessary in Colombia and Latin America, in order to heighten awareness of contemporary MRSA clones circulating in specific countries and institutions and mitigate their impact on the design of effective strategies for controlling MRSA transmission. Unfortunately, individual hospitals within Colombia have typically implemented their own infection control policies. Recently, however, the National Institute of Health has started implementing a national surveillance program for healthcare-related infections, including antimicrobial consumption and resistance patterns, within hospitals.

Supporting Information
Table S1 Results of MLST in representative strains of major spa types.