Molecular Characterization of Staphylococcus aureus from Patients with Surgical Site Infections at Mulago Hospital in Kampala, Uganda

Background The prevalence of Methicillin resistant Staphylococcus aureus (MRSA) is progressively increasing globally with significant regional variation. Understanding the Staphylococcus aureus lineages is crucial in controlling nosocomial infections. Recent studies on S. aureus in Uganda have revealed an escalating burden of MRSA. However, the S. aureus genotypes circulating among patients are not known. Here, we report S. aureus lineages circulating in patients with surgical site infections (SSI) at Mulago National hospital, Kampala, Uganda. Methods A cross-sectional study involving 314 patients with SSI at Mulago National Hospital was conducted from September 2011 to April 2012. Pus swabs from the patients’ SSI were processed using standard microbiological procedures. Methicillin sensitive Staphylococcus aureus (MSSA) and MRSA were identified using phenotypic tests and confirmed by PCR-detection of the nuc and mecA genes, respectively. SCCmec genotypes were determined among MRSA isolates using multiplex PCR. Furthermore, to determine lineages, spa sequence based-genotyping was performed on all S. aureus isolates. Results Of the 314 patients with SSI, S. aureus accounted for 20.4% (64/314), of which 37.5% (24/64) were MRSA. The predominant SCCmec types were type V (33.3%, 8/24) and type I (16.7%, 4/24). The predominant spa lineages were t645 (17.2%, 11/64) and t4353 (15.6%, 10/64), and these were found to be clonally circulating in all the surgical wards. On the other hand, lineages t064, t355, and t4609 were confined to the obstetrics and gynecology wards. A new spa type (t10277) was identified from MSSA isolate. On multivariate logistic regression analysis, cancer and inducible clindamycin resistance remained as independent predictors of MRSA-SSI. Conclusion SCCmec types I and V are the most prevalent MRSA mecA types from the patients’ SSI. The predominant spa lineages (t645 and t4353) are clonally circulating in all the surgical wards, calling for strengthening of infection control practices at Mulago National Hospital.


Introduction
Staphylococcus aureus is a leading causative agent of surgical site infections (SSI) worldwide [1]. S. aureus carriage is the most important factor predisposing colonized individuals to subsequent SSI; other factors are longer duration of hospital stay, HIV infection, cancer, and opsonic defects [2,3]. Surveillance data have shown that in hospital settings methicillin sensitive Staphylococcus aureus (MSSA) tends to evolve into methicillin resistant Staphylococcus aureus (MRSA) [4,5]. Globally the prevalence of MRSA is progressively increasing with significant regional variation [4,5]. Data in most African countries are scarce; in few surveys the prevalence of MRSA has been found to range from as low as 2% to as high as 41% [6,7,8,9,10]. In Uganda, previous studies at Mulago National Hospital showed that 28.7% of the SSI are due to S. aureus, and of these 31.5% are due to MRSA [11].
The association of MRSA with therapeutic challenges, complications, deaths and cost related to longer hospital stay compared with MSSA has been widely documented [12,13,14]. More importantly, the multidrug resistant hospital-acquired MRSA (HA-MRSA) strains and their intrinsic resistance to beta-lactam antibiotics confer limited treatment options to the most available and less costly antibiotics in developing countries [11,15].
Molecular approaches are of paramount importance in showing clonality and spreading patterns of S. aureus strains in hospital settings; however, variation exists about their effectiveness, cost and general applicability [16,17]. PCR based approaches for characterizing Staphylococcal chromosomal cassettes mec (SCCmec) types and determining sequence polymorphism in the variable X region of the Staphylococcal protein a (spa) have shown to be relatively less expensive, easier, less time consuming and have clear discriminatory power in comparison with other approaches, such as multi locus sequence typing (MLST) and pulse field gel electrophoresis (PFGE) [18,19,20].
SCCmec types can correlate with the classic HA-MRSA (SCCmec types I-III) and community associated MRSA (CA-MRSA) (SCCmec types IV and V) [21,22,23]. Furthermore, the pvl gene, which is a virulence factor for severe staphylococcal infections, has also been associated with CA-MRSA [24]. Spa sequence typing can show clonality among S. aureus and has portability of data with absolute reproducibility which allows internet-based type assignment and worldwide comparison of isolates [25,26].
Data showing S. aureus genotypes circulating among patients with SSI at Mulago National Hospital to guide infection control surveillance remain scanty. Furthermore, whether there is clonal spread of the predominant genotypes is also not known. Using spa sequence typing and SCCmec genotyping, we present S. aureus lineages in patients with SSI at Mulago National Hospital in Kampala, Uganda.

Study Design and Sampling Procedures
This cross-sectional study was conducted from September 2011 to April 2012 at Mulago National Hospital, the largest hospital located in Kampala, Uganda, with an estimated 1,500 beds. Using the Kish and Lisle formula with prevalence of 28.7%, a sample size of 314 patients with clinical SSI from obstetrics & gynecology, general surgery and orthopedic wards were enrolled [11]. Consenting patients with SSI occurring within 30 days after the operative procedure or within one year if orthopedic implant is in situ were included in the study, while surgical patients with community acquired pyogenic infections such as abscess, furuncle and carbuncle; patients with infection of an episiotomy; and patients with open fractures were excluded.

Data Collection and Laboratory Procedure
Demographic and clinical characteristics from consenting patients were collected using structured pretested questionnaire and patients' files. These included age, sex, ward of admission, operation type, extent of SSI, presence of cancer, HIV serostatus, and length of hospital stay.
The infected site was cleaned using normal saline and sterile gauze, then pus was collected by a sterile cotton swab in moist Amies transport media [27]. Two pus samples were collected from each patient. A nasal swab was also taken from each patient for screening of S. aureus carriage. The samples were transported within two hours of collection to the clinical laboratories of the Department of Medical Microbiology, Makerere University College of Health Sciences for processing. Figure S1 shows the study flow chart.
The first pus swab was used to make gram stain smears while the second one was inoculated onto Blood agar and Mannitol-salt agar and incubated at 35-37uC for 24 hours for recovery of staphylococci. Isolates were initially identified using standard microbiological procedures based on colony morphology on Blood agar which are creamy to golden yellow in color with or without hemolysis, and on Mannitol-salt agar by formation of yellow zone around colonies. For phenotypic confirmation of S. aureus, the catalase test was done followed by slide and tube coagulase tests, as well as DNAse test.
Apart from conventional methods, isolates confirmation and drug susceptibility testing were done using the Phoenix Automated instrument (Becton-Dickson, Sparks Maryland) as per manufacturer's instruction.

Molecular Assays
All PCR primers used in this study are described in Table S1. i) Detection of the nuc gene. S. aureus chromosomal DNA was extracted as previously described [21] and species confirmed through PCR-amplification and identification of a 270 bp amplicon of the S. aureus thermonuclease (nuc) gene [29,30]. The PCR program comprised of initial denaturation at 94uC for 5 minutes followed by 30 cycles of denaturation at 94uC for 45 seconds, primer annealing at 50uC for 45 seconds, extension at 72uC for 45 seconds, and a final extension at 72uC for 10 minutes. The amplicons were analyzed by electrophoresis on a 2% agarose gel (Table S1). Images were captured and retrieved using the Bioimager (UVP, Upland, California, USA).
ii) Detection of the mecA and pvl genes. Confirmation of MRSA was based on PCR-detection of the mecA gene (usually the gold standard) through amplification and identification of a 162 bp amplicon upon agarose gel electrophoresis. Presence of the pvl gene was also determined by PCR-amplification and detection of a 433 bp region overlapping the lukS-PV and lukF-PV genes. Both mecA and pvl genes were detected through a multiplex PCR as previously described [31]. The PCR program comprised of initial denaturation at 94uC for 5 minutes followed by 30 cycles of denaturation at 94uC for 30 seconds, primer annealing at 53uC for 30 seconds, extension at 72uC for 45 seconds, and a final extension at 72uC for 10 minutes. The amplicons were analyzed by electrophoresis on a 2% agarose gel (Table S1).
iii) SCCmec genotyping. The multiplex PCR we followed in typing SCCmec types I-V was described previously [22]. However, for improved efficiency, in this study we introduced minor modifications in the protocol such as using two primer pairs in each case (i.e. ccrC and 5R; ab and IS1272). The PCR program comprised of initial denaturation at 94uC for 5 minutes followed by 30 cycles of denaturation at 94uC for 45 seconds, primer annealing at 50uC for 45 seconds, extension at 72uC for 45 seconds, and a final extension at 72uC for 10 minutes. The amplicons were analyzed by electrophoresis on a 2% agarose gel (Table S1). Following PCR, interpretation of amplicons was based on number and size of band(s) for the four target amplicons upon agarose gel electrophoresis; where no band was observed the respective isolate was regarded as ''non-typeable''. Repeated testing and optimization of PCR conditions were done in which the results of non-typeable isolates results were reproducible (i.e. no amplification following repeated PCRs). iv) spa genotyping and sequence analysis. To determine the genetic lineages of S. aureus, the x-region of the spa gene (200 bp to 400 bp) was amplified by PCR using the method established before [25]. The PCR program comprised of initial denaturation at 94uC for 5 minutes followed by 30 cycles of denaturation at 94uC for 45 seconds, primer annealing at 50uC for 45 seconds, extension at 72uC for 45 seconds, and a final extension at 72uC for 10 minutes. The amplicons were analyzed by electrophoresis on a 2% agarose gel (Table S1). The PCR products were purified with the QIAquick PCR purification kit (Qiagen, Hilden, Germany) as per the manufacturer's instructions, and both strands were sequenced by ACGT (Wheeling, IL, USA) using the same forward and reverse primers (Table S1). To determine spa types, sequences were submitted to the spaTyper data base (http://fortinbras.us/cgi-bin/spaTyper/spaTyper.pl) and spa lineages matching to the query sequence determined. The data was also matched with the Ridom SpaServer (Ridom GmbH, Wurzburg, Germany) (http://spa.ridom.de/). Chromatograms for unique sequences (i.e. lacking hits) were submitted to the Ridom SpaServer (http://spa.ridom.de/) to assess novelty of lineages.

Quality Control
Standard aseptic sample collection and processing measures were strictly adhered. For phenotypic identification of S. aureus; S. aureus ATCC 25923 and S. epidermidis ATCC 12228 were used as positive and negative controls, respectively. For genotyping, S. aureus ATCC 29213 (mecA-negative, PVL-negative), and ATCC 43300 (mecA-positive, PVL-negative) were used.

Data Analysis
Variables from demographic, clinical and laboratory data were entered into excel spreadsheet, cleaned and exported to STATA software version 11 (College Station, Texas, USA) for analysis according to the study objectives. Continuous variables were summarized as mean (6 standard deviation). Categorical variables were described as proportion and were analyzed to compare the significance of difference in distribution by using Chi-square test or Fischer's exact test where appropriate. All variables with association using univariate logistic regression were subjected to multivariate logistic regression analysis. Odds ratio with respective 95% confidence interval was used to measure the strength of association between the surgical site infection and predictive variables. Variables with p-value less than 0.05 were considered as independent predictors of surgical site infections.  HS 1080). A written informed consent (or assent) was obtained from each patient/caretaker. All patient information was kept confidential and anonymous using codes. Results for antimicrobial susceptibility testing were promptly reported to the attending physician for patient care.

Results
Three hundred and fourteen (314) (Table 1). All S. aureus isolates were susceptible to vancomycin. Two MRSA isolates were misidentified as MSSA by the disc diffusion method using oxacillin (1 mg) and cefoxitine (30 mg) discs, but were confirmed to be MRSA upon mecA-PCR. Their respective zones of inhibition were 22 mm and 23 mm for cefoxitin, and 13 mm for oxacillin for both.

Genetic Lineages of S. aureus based on spa Typing and their Distribution in Different Wards
A total of 14 spa lineages were determined. One third of these fell in two major lineages namely t645 and t4353, which accounted for 17.2% (11/64) and 15.6% (10/64), respectively. The major lineages were found to be clonally circulating in all the surgical wards whereas lineages t064, t355, and t4609 were confined to obstetrics and gynecology wards. A new spa type namely t10277 was also identified in this study from MSSA isolate ( Table 2). The association of these spa types with MSSA and MRSA are shown in Table S2. Apparently, 11 isolates had sequent repeats with no identity in the spa server (i.e. novel clones) and the spa gene was not detected in four isolates on sequencing, despite a repeated reactions.
Additionally, of the 64 patients with S. aureus SSI, six were colonized by S. aureus in the nose and only one was concurrently infected and colonized with the same MRSA strain (SCCmec V and spa type t657). Thus, the overall nasal colonization rate with S. aureus was 9.4% (6/64) while nasal colonization rate with the same strain causing SSI was 1.6% (1/64).

Discussion
The prevalence of S. aureus in patients with SSI in this study (20.4%) was lower than that found in other studies (28.7% and 28.6%) [9,11]. However, the proportion of MRSA among S. aureus was higher (37.5%) than previously reported from the same setting (31.5%) [11] and 4.8%-29.6%, 25%, and 18.8% from other studies in eight African countries, Jinja-Uganda and Mwanza-Tanzania, respectively [6,9,32]. The variability in prevalence may be influenced by infection control practices in different settings.
The finding that MRSA is associated with multi drug resistance compared to MSSA in this study have also been reported in other studies [6,33,34]. Similar to another study in the same region [35], S. aureus exhibiting phenomenal inducible clindamycin resistance (D-test positive) accounted for 17.2%, more were found in MRSA compared to MSSA isolates (p = 0.002). This implies that both clindamycin and erythromycin cannot be used in these patients.
In this study, the proportion of females with SSI was higher compared to other studies [9,14,32], whereby females accounted for 40% to 60.0%. This was because more than half of patients were from obstetrics and gynecology wards. Similar to other studies [9,14,32], the median age was variably ranging from 30 to 60 years.   On univariate analysis, there was a statistical significant association between MRSA-SSI with infections involving organ, cancer, and inducible clindamycin resistance (p = 0.005, 0.022, and 0.002, respectively) in comparison with MSSA-SSI. Furthermore, on multivariate logistic regression analysis, cancer and inducible clindamycin resistance remained as independent predictors of MRSA-SSI. Other studies have reported that long duration of hospital stay and presence of co-morbid conditions have been found to be more associated with MRSA-SSI [1,9,14].
Of the 14 spa lineages found in this study, one third fell among two major lineages (t645 and t4353) and were found to be clonally circulating in all the surgical wards; on the other hand, t064, t355, and t4609 were confined to obstetrics and gynecology wards. This calls for strengthening of infection control practices at Mulago National Hospital to prevent this clonal spread of S. aureus lineages. There seems to be diversity between spa types in different regions and different countries. Indeed none of the spa lineages in the index study were found in two studies in the USA and UK [16,36] whereas lineages t037, t064, t084, t127, t355, and t3772 found in this study were similar to those previously reported from Nigeria [34]. t064, also found in this study was reported in Mwanza, Tanzania [15]. The variability in spa lineages across regions emphasizes the need for region wide infection control surveillance.
In this study, SCCmec type V was the most predominant among MRSA, followed by SCCmec types I, IV and III. The predominance of SCCmec type V was previously reported among isolates from the burns units at Mulago National Hospital [37] and in another study in Nigeria [34]. However, in these other studies SCCmec types I and III were also predominant. SCCmec types I, II, and III are usually ascribed to HA-MRSA isolates whereas SCCmec types IV and V are ascribed to CA-MRSA (which accounted for 29.1% and 45.8%, respectively, of the MRSA isolates in the current study) [23]. Indeed, the findings from Uganda and Nigeria may suggest presence of mixed CA-MRSA and HA-MRSA genotypes in hospital settings, supporting the notion that there is a changing epidemiology reflected by community associated SCCmec genotypes being now more associated with hospital infections [38]. Nevertheless, based on the protocol by Boye et al (2007) [22], a quarter of the isolates in this study were non-typeable despite optimization of laboratory methods suggesting they could belong to SCCmec types other than I-V (i.e. SCCmec types VI-VIII).
Of the 64 patients with S. aureus SSI, six were colonized by S. aureus in the nose and only one was concurrently infected and colonized with the same MRSA strain (SCCmec V and spa type t657). This suggests an external source of infection for the MRSA-SSI, either from healthcare workers, environment or other sources.

Conclusions
SCCmec types V and I are the most prevalent MRSA genotypes. S. aureus spa lineages t645 and t4353 which were predominant are clonally circulating in all surgical wards, calling for strengthening of infection control practices at Mulago National Hospital. Studies to determine the sources of these lineages would be of interest, as well as studies comparing these isolates with other continental lineages.