INTRODUCTION
Staphylococcus aureus is a major human pathogen, both in adults and children [Reference David and Daum1, Reference Gutierrez2]. The spectrum of S. aureus disease ranges from non-invasive manifestations such as skin and soft tissue infection (SSTI), to serious invasive disease such as pneumonia and endocarditis [Reference Boucher and Corey3]. Over the past two decades, the incidence of both invasive and non-invasive S. aureus disease has increased in several settings, including Europe, North America and New Zealand [Reference David and Daum1, Reference Klevens4–Reference de Kraker7]. In many countries, particularly the USA, this increase has been driven by the unprecedented rise in infections caused by a predominant strain of community-associated methicillin-resistant S. aureus (CA-MRSA) [Reference Moran8].
New Zealand has a high incidence of both invasive and non-invasive S. aureus disease [Reference Williamson5, Reference Muttaiyah9–Reference Hill11]. Moreover, similar to reports from North America, the incidence of community-associated paediatric SSTI is increasing in New Zealand [Reference Williamson5, Reference O'Sullivan, Baker and Zhang12]. However, in contrast to North America, most S. aureus infections in New Zealand are caused by methicillin-susceptible S. aureus (MSSA) strains, rather than CA-MRSA [Reference Hill10, Reference Hill11].
Clinical isolates of S. aureus harbour an array of virulence factors involved in adhesion, invasion and immune evasion [Reference David and Daum1, Reference Gordon and Lowy13, Reference Tong14]. In particular, the Panton–Valentine leucocidin toxin (PVL) has been extensively studied regarding its role in disease pathogenesis and severity [Reference Shallcross15–Reference Gillet18]. PVL is a bi-component pore-forming cytotoxin and has been implicated as an important virulence factor in both invasive and non-invasive S. aureus disease [Reference Muttaiyah9, Reference Gillet18]. However, a recent meta-analysis found that, in comparison with isolates causing SSTI, the lukF-PV and lukS-PV genes encoding PVL were less commonly identified in S. aureus isolates causing invasive disease [Reference Shallcross15]. Moreover, the presence of the lukF-PV and lukS-PV genes in invasive S. aureus isolates was not shown to correlate with adverse clinical outcomes [Reference Shallcross15]. Similarly, a recent Australian study also found no difference in mortality between patients with ‘PVL-positive’ and ‘PVL-negative’ invasive S. aureus infections [Reference Wehrhahn19]. In addition, a previous study in our setting found that the prevalence of the lukF-PV and lukS-PV genes in nasal carriage MSSA isolates was the same as the prevalence of the lukF-PV and lukS-PV genes in disease-causing MSSA isolates (31% vs. 37%, respectively; P = 0·33), suggesting that PVL is not the primary determinant of S. aureus disease severity [Reference Muttaiyah9]. To date, however, few studies have systematically assessed and compared molecular determinants of virulence, other than PVL, between S. aureus isolates causing invasive and non-invasive disease.
Accordingly, we sought to (i) describe the incidence and demographics of invasive S. aureus disease in children admitted to our hospital between 2007 and 2010, and (ii) compare the CCs and virulence gene repertoire (including PVL), of S. aureus causing invasive and non-invasive disease in these patients.
METHODS
Setting, patients and isolates
Starship Children's Hospital (SCH) in Auckland, New Zealand is a tertiary-level university-affiliated institution, serving a population of ∼500 000 inhabitants, within a larger metropolitan population of 1·46 million. We performed a retrospective, cross-sectional analysis of all children (aged < 15 years) admitted to SCH with S. aureus infections over the 4-year period from January 2007 and December 2010. All children with S. aureus isolated from a clinical specimen were identified from the laboratory database in the Department of Microbiology, Auckland City Hospital. A unique S. aureus infection was defined as the first positive S. aureus culture taken from the same patient within a 30-day period. All S. aureus isolates were phenotypically identified using standard laboratory methods. Antimicrobial susceptibility testing of S. aureus isolates was performed by agar dilution, and results were interpreted according to Clinical and Laboratory Standards Institute recommendations [20].
Data collection and case definition
Using the hospital information database, demographic information was obtained about each patient who cultured S. aureus from a clinical specimen. The following information was recorded: age, gender, ethnicity, and discharge diagnoses associated with each hospital admission, based on the International Classification of Disease, Tenth Revision (ICD-10) codes [21].
Similar to previously described methodology [Reference Tracy22], a list of ICD-10 codes was developed for S. aureus-related clinical syndromes (see online Supplementary Table S1). These ICD-10 codes were then grouped into the following broad categories for analysis: (i) skin and soft tissue infection (SSTI), (ii) musculoskeletal infection, (iii) respiratory infection, (iv) endovascular infection, (v) central nervous system infection and (vi) bacteremia, site not specified. For cases that had ⩾1 of these ICD-10 codes associated with a S. aureus culture, the principal discharge diagnosis was regarded as the representative clinical syndrome. A case was described as community-onset if S. aureus was isolated from a patient within 48 h of hospital admission.
All cases of S. aureus bloodstream infection were classified as invasive infections. In addition, other invasive S. aureus infections were defined by the isolation of S. aureus from a sterile body site (as defined by the Centers for Disease Control and Prevention Active Bacterial Core surveillance program [Reference Klevens4]), plus ⩾1 ICD-10 discharge codes associated with an appropriate clinical syndrome (Supplementary Table S1).
Molecular characterization and comparison of invasive vs. non-invasive S. aureus isolates
Detailed molecular characterization of invasive S. aureus isolates was performed by DNA microarray analysis (StaphType, Clondiag, Germany) using previously described methods [Reference Monecke23]. In brief, this assay detects 334 target sequences in the S. aureus genome, corresponding to 186 genes and their allelic variants. This allows for: (i) broad classification of S. aureus into major clonal complexes (CCs) [Reference Monecke, Slickers and Ehricht24], (ii) classification of the SCCmec complex into distinct allotypes [Reference Monecke25], and (iii) detection of genes associated with antimicrobial resistance, virulence and adhesion [Reference Monecke, Slickers and Ehricht24]. In order to compare molecular features of invasive and non-invasive isolates, DNA microarray analysis was also performed on a sample of S. aureus isolates obtained from paediatric patients attending SCH with uncomplicated, non-invasive S. aureus SSTI. These isolates were obtained during the same study time period (between February 2008 and April 2008), and were collected as part of a previous study conducted in our department [Reference Muttaiyah9].
Statistical analysis
Age-standardized incidence rates for children presenting with community-onset invasive S. aureus infections were calculated using denominator population information from the 2006 New Zealand census, and from projected population data for the Auckland region [26]. Rates were calculated for the following ethnicities: European, Māori (Indigenous New Zealander), Pacific Peoples, and other ethnicities. Categorical variables were compared using the χ 2 or Fisher's exact tests, and non-parametric variables were compared using the Kruskal–Wallis analysis of variance test. Two-dimensional multi-dimensional scaling (MDS) was used to visualize S. aureus subgroups following formation of a distance matrix derived from the DNA array using PRIMER v. 6.1.15 (Primer-E, UK). The null hypothesis that there was no difference in virulence genes between invasive and non-invasive S. aureus isolates was tested using permutational analysis of variance (PERMANOVA) [Reference Anderson27]. In addition, the interaction between CCs, virulence genes and invasive vs. non-invasive status was also assessed using PERMANOVA. Statistical analysis was performed using GraphPad Prism version 5.02 (GraphPad Software, USA), PERMANOVA+ version 1.0.2 (Primer-E, UK) or R version 3.0.1 (R Foundation, Austria). A two-tailed P value of < 0·05 was considered statistically significant.
Ethical approval
The Northern X Ethics Committee, New Zealand, granted ethical approval for this study.
RESULTS
Incidence and clinical features of co-invasive S. aureus infections
Between January 2007 and December 2010, 2329 children had S. aureus isolated from a clinical specimen plus ⩾1 ICD-10 discharge code(s) associated with a S. aureus-related clinical syndrome. Of these 2329 children, 163 (163/2329, 7%) were classified as having community-onset invasive S. aureus infection (Table 1). The average annual incidence of invasive community-onset S. aureus disease in children during the study period was 52/100 000 population [95% confidence interval (CI) 44–60/100 000]. When stratified by age, the highest incidence was in the <1 year age group (83/100 000 population, 95% CI 52–130/100 000), and when stratified by ethnicity, the highest incidence was in Māori children (132/100 000 population, 95% CI 98–176/100 000) (Fig. 1). The commonest clinical syndrome associated with invasive S. aureus infection was musculoskeletal infection (91/163, 56%), followed by bacteraemia, site not identified (37/163, 23%), respiratory infection (27/163, 17%), endovascular infection (7/163; 4%) and central nervous system infection (1/163, 1%) (Table 1).
Fig. 1. Incidence of community-onset invasive Staphylococcus aureus infection in children admitted to Starship Children's Hospital between 2007 and 2010, stratified by (a) ethnicity and (b) age.
Table 1. Clinical characteristics of children with community-onset invasive Staphylococcus aureus infections admitted to Starship Children's Hospital, Auckland, New Zealand, 2007–2010
IQR, Interquartile; CNS, central nervous system.
Isolates and comparison of molecular features of invasive and non-invasive S. aureus isolates
Of the 163 children with invasive S. aureus disease, 19 children (19/163, 12%) had isolates that were methicillin-resistant. Of the 163 invasive S. aureus disease episodes, isolates from 72 episodes were available from stocked cultures for molecular characterization (68 MSSA, 4 MRSA). Of these 72 isolates, 54 (75%) were associated with musculoskeletal infection, 10 (14%) with respiratory infection, six (8%) with bacteremia and two (3%) with endovascular infection. In order to identify possible genotypic associations with invasiveness, the DNA microarray profiles of these 72 isolates were compared to the DNA microarray profiles of a convenience sample of 54 non-invasive S. aureus SSTI isolates from children collected during the study period (50 MSSA, 4 MRSA).
Based on DNA microarray results, a total of 18 different CCs were detected among the invasive and non-invasive isolates (Fig. 2). The DNA microarray was unable to resolve the CCs of two isolates. The three predominant CCs causing SSTI or invasive disease during the study period were CC1 (45/126, 36%), CC121 MSSA (19/126, 15%) and CC30 MSSA (17/126, 13%). The lukF-PV and lukS-PV genes were found in isolates from eight CCs: CC25 (3/3, 100%), CC88 (2/2, 100%), CC121 (16/19, 84%), CC30 (12/17, 71%), CC1 (34/45, 67%), CC45 (2/7, 29%), CC59 (1/4, 25%) and CC5 (1/6, 17%).
Fig. 2. Distribution of clonal complexes (CC) (as determined by DNA microarray) in invasive and non-invasive isolates of Staphylococcus aureus.
When tested by univariate analyses, there was no significant difference in the prevalence of virulence, regulatory or adhesion genes between invasive and non-invasive S. aureus isolates (Table 2). Of note, there was no difference in the detection of the lukF-PV and lukS-PV genes between invasive and non-invasive S. aureus isolates (42/72, 58% vs. 29/54, 54%; P = 0·72). The commonest agr group detected in both invasive and non-invasive isolates was agr group III (54% and 50%, respectively). On PERMANOVA testing, there were no significant differences in the virulence gene repertoire between invasive and non-invasive strains (pseudo-F = 1·3337, P = 0·1731) (Fig. 3). Figure 4 shows the MDS map of 126 S. aureus isolates differentiated on the basis of their virulence gene repertoire. This MDS map demonstrates that, for example, the virulence gene repertoire of CC1 isolates (green triangles) is distinct to the virulence gene repertoire of CC30 isolates (black squares). Explored statistically, a significant interaction was observed between CCs and invasive status (pseudo-F = 2·0625, P = 0·0001), suggesting that distinct combinations of virulence genes may be present in invasive isolates within specific CCs (Fig. 4). On pairwise testing between invasive and non-invasive strains, a significant difference between invasive and non-invasive status was observed within three CCs: CC1 (t = 2·0853, P < 0·001), CC30 (t = 3·2966, P = 0·007), and CC121 (t = 2·1472, P = 0·02) (Fig. 4). However, our sample size within CCs was insufficient to conclusively determine which combination(s) of genes were responsible for these statistical differences.
Fig. 3 [colour online]. Non-metric multidimensional scaling of virulence genes in invasive and non-invasive Staphylococcus aureus isolates based on Euclidean distances.
Fig. 4. Non-metric multidimensional scaling of virulence genes in Staphylococcus aureus isolates according to clonal complex (CC) (as determined by DNA microarray) based on Euclidean distances.
Table 2. Prevalence and comparison of regulatory and virulence genes in invasive and non-invasive Staphylococcus aureus isolates from children admitted to Starship Children's Hospital, Auckland, New Zealand, 2007–2010
MSCRAMMS, Microbial surface components recognizing adhesive matrix molecules; ACME, arginine catabolic mobile element.
All values are numbers of isolates (%) unless otherwise specified.
* Fisher's exact test.
DISCUSSION
In this study, we measured the incidence of community-onset invasive S. aureus disease in children presenting to our hospital between 2007 and 2010, and systematically compared the genotypic characteristics of invasive and non-invasive S. aureus isolates. Using univariate testing, we found no difference in the clonal distribution of S. aureus strains or prevalence of tested virulence factors between invasive and non-invasive isolates. However, we observed a significant interaction between CC and invasive status, with three CCs (CC1, CC121, CC30) demonstrating a significant association with invasive S. aureus infections. This finding is in keeping with other studies that have also observed an interaction between specific S. aureus CCs and clinical syndromes. For example, one previous study found a strong association between CC30 MSSA and infective endocarditis [Reference Fowler28] and another recent study found an association between distinct subgroups of CC121 S. aureus and disease type [Reference Kurt29]. Although we were unable to determine which genes were responsible for the association between CC and invasiveness, our findings suggest that there may be differences in invasive and non-invasive S. aureus disease according to phylogenetic grouping. Notably, we found no difference in the prevalence of lukF-PV and lukS-PV genes between S. aureus isolates causing invasive disease and those S. aureus isolates causing uncomplicated SSTI. This finding is similar to other recent studies that also found no difference in disease type based on the presence of PVL genes [Reference Muttaiyah9, Reference Wehrhahn19]. Interestingly, however, we found an unexpectedly high prevalence of lukF-PV and lukS-PV genes in both invasive and non-invasive MSSA isolates (57% and 50%, respectively). This rate of PVL-positive MSSA is higher than that reported from several previous studies [Reference Muttaiyah9, Reference Munckhof30–Reference Orscheln32], but is comparable to a recent African study, in which the lukF-PV and lukS-PV genes were detected in 130/228 (57%) clinical MSSA isolates [Reference Breurec33]. In addition, our rate of PVL-positive MSSA is higher than that reported from a previous study in New Zealand, which detected the lukF-PV and lukS-PV genes in 37% of disease-causing MSSA isolates [Reference Muttaiyah9]. The reasons for the high rate of PVL-positive MSSA in our study are unclear, but may be partly due to the fact that we only assessed MSSA isolates from paediatric patients. The age-specific incidence of infections caused by PVL-producing S. aureus strains has been previously described, with younger patients having a disproportionately higher rate of infections due to PVL-producing S. aureus strains [Reference Munckhof30]. Although this association may be partially mediated by age-related immunological protection, supportive data for this hypothesis remain inconclusive [Reference Croze34, Reference Hermos, Yoong and Pier35].
Our study also provides valuable information on the genotypes of MSSA strains circulating in New Zealand. Despite the fact that about 90% of S. aureus infections in New Zealand are caused by MSSA strains [Reference Hill10, Reference Hill11], only one previous study has provided data on the molecular epidemiology of MSSA strains circulating in our region [Reference Muttaiyah9]. Based on DNA microarray data, we found a diverse range of MSSA CCs causing disease, although three MSSA strain types predominated (CC1, CC121, CC30) and together accounted for two-thirds of all MSSA strains identified. Interestingly, in one child with osteomyelitis, we detected a MSSA isolate belonging to CC398. Although CC398 MRSA has recently been described as a cause of CA-MRSA infection in New Zealand [Reference Williamson36], to our knowledge, CC398 MSSA has not previously been reported in our country.
Finally, we observed significant demographic variation in the incidence of community-onset invasive S. aureus disease among children in our setting. Ethnic variation in invasive S. aureus disease has been described previously, most notably in North America, where in one study, the incidence of invasive MRSA infections was more than four times higher in black infants compared to white children from the same population [Reference Klevens4]. In addition, Gutierrez et al. found that among children in California, black children were 1·5 times more likely to be hospitalized with staphylococcal infection than white children [Reference Gutierrez2]. The reasons for the markedly high disease burden in Māori and Pacific Island children in New Zealand are unclear, but are likely to be due, in part, to a combination of societal, environmental and host factors, including barriers in accessing healthcare, household crowding, hygiene and nutrition. Future studies should assess modifiable risk factors that may reduce this serious disease burden and ethnic disparity.
There are a number of limitations to our study. In particular, the retrospective nature of analysis limited the amount of data we could collect, particularly information relating to duration of preceding disease course, prior antibiotic usage, medical co-morbidities, and clinical outcomes. In addition, the DNA microarray assay we used for genotypic typing of staphylococcal CCs does not provide the same degree of resolution as other molecular typing methods, such as multi-locus sequence typing or spa typing. It is, therefore, possible that there may be additional epidemiological or molecular associations with distinct S. aureus sublineages within CCs that we were unable to detect. Finally, the DNA microarray we used only detects the presence of commonly studied virulence genes – it is not a complete representation of virulence genes within S. aureus, nor does it quantify virulence gene expression.
In summary, our study provides valuable information on the clinical and molecular epidemiology of invasive S. aureus disease in New Zealand children. We observed notable sociodemographic variation in the incidence of S. aureus disease in our setting, with rates highest in Māori and Pacific Island children. Most disease was caused by MSSA strains, which were genetically diverse. Based on DNA microarray data, we identified an interaction between invasive disease and distinct S. aureus CCs. Future research should include a strong focus on identifying the host and environmental factors that, along with other potential organism virulence factors, are contributing to the patterns of invasive S. aureus disease observed in New Zealand.
SUPPLEMENTARY MATERIAL
For supplementary material accompanying this paper visit http://dx.doi.org/10.1017/S0950268814000053.
ACKNOWLEDGEMENTS
This study was supported by internal funding. D.A.W. is a Clinical Research Training Fellow of the Health Research Council of New Zealand.
DECLARATION OF INTEREST
None.
