Antimicrobial resistance in coagulase-positive staphylococci isolated from companion animals in Australia: A one year study

Methicillin-resistant coagulase-positive staphylococci (CoPS) have become increasingly recognised as opportunistic pathogens that limit therapeutic options in companion animals. The frequency of methicillin resistance amongst clinical isolates on an Australia-wide level is unknown. This study determined antimicrobial susceptibility patterns for CoPS isolated from clinical infections in companion animals (dogs, cats and horses) as part of the first nation-wide survey on antimicrobial resistance in animal pathogens in Australia for a one-year period (January 2013 to January 2014). Clinical Staphylococcus spp. isolates (n = 888) obtained from 22 veterinary diagnostic laboratories were identified by MALDI-TOF mass spectrometry and subjected to antimicrobial susceptibility testing for 16 antimicrobials, representing 12 antimicrobial classes. Potential risk factors associated with methicillin resistance in Staphylococcus pseudintermedius isolates from dogs were analysed based on demographic factors and clinical history, including gender, age, previous antimicrobial treatment, chronic and/or recurrent diseases and site of infections. The most commonly identified CoPS were S. pseudintermedius (70.8%; dogs n = 616, cats n = 13) and S. aureus (13.2%, horses n = 53, dogs n = 47 and cats n = 17). Overall, the frequency of methicillin resistance among S. pseudintermedius (MRSP) and S. aureus (MRSA) was 11.8% and 12.8%, respectively. MRSP isolates were strongly associated with resistance to fluoroquinolones (OR 287; 95%CI 91.2–1144.8) and clindamycin (OR 105.2, 95%CI 48.5–231.9). MRSA isolates from dogs and cats were also more likely to be resistant to fluoroquinolones (OR 5.4, 95%CI 0.6–252.1), whereas MRSA from horses were more likely to be resistant to rifampicin. In multivariate analysis, MRSP-positive status was significantly associated with particular infection sites, including surgical (OR 8.8; 95%CI 3.74–20.7), and skin and soft tissue (OR 3.9; 95%CI 1.97–7.51). S. pseudintermedius isolated from dogs with surgical site infections were three times more likely to be methicillin-resistant if cases had received prior antimicrobial treatment. Whilst the survey results indicate the proportion of CoPS obtained from Australian companion animals that are methicillin-resistant is currently moderate, the identified risk factors suggest that it could rapidly increase without adequate biosecurity and infection control procedures in veterinary practice.

Methicillin-resistant coagulase-positive staphylococci (CoPS) have become increasingly recognised as opportunistic pathogens that limit therapeutic options in companion animals. The frequency of methicillin resistance amongst clinical isolates on an Australia-wide level is unknown. This study determined antimicrobial susceptibility patterns for CoPS isolated from clinical infections in companion animals (dogs, cats and horses) as part of the first nation-wide survey on antimicrobial resistance in animal pathogens in Australia for a oneyear period (January 2013 to January 2014). Clinical Staphylococcus spp. isolates (n = 888) obtained from 22 veterinary diagnostic laboratories were identified by MALDI-TOF mass spectrometry and subjected to antimicrobial susceptibility testing for 16 antimicrobials, representing 12 antimicrobial classes. Potential risk factors associated with methicillin resistance in Staphylococcus pseudintermedius isolates from dogs were analysed based on demographic factors and clinical history, including gender, age, previous antimicrobial treatment, chronic and/or recurrent diseases and site of infections. The most commonly identified CoPS were S. pseudintermedius (70.8%; dogs n = 616, cats n = 13) and S. aureus (13.2%, horses n = 53, dogs n = 47 and cats n = 17). Overall, the frequency of methicillin resistance among S. pseudintermedius (MRSP) and S. aureus (MRSA) was 11.8% and 12.8%, respectively. MRSP isolates were strongly associated with resistance to fluoroquinolones (OR 287; 95%CI 91. 2-1144.8) and clindamycin (OR 105.2, 95%CI 48.5-231.9). MRSA isolates from dogs and cats were also more likely to be resistant to fluoroquinolones (OR 5.4, 95%CI 0.6-252.1), whereas MRSA from horses were more likely to be resistant to rifampicin. In multivariate analysis, MRSP-positive status was significantly associated with particular infection sites, including surgical (OR 8.8; 95%CI 3.74-20.7), and skin and soft tissue (OR 3.9; 95%CI 1.97-7.51). S. pseudintermedius isolated from dogs with surgical site infections were three times more likely to be methicillin-resistant if cases had received prior antimicrobial treatment. Whilst the survey results indicate the proportion of CoPS obtained from Australian companion animals that are methicillin-resistant is currently moderate, the Introduction Coagulase-positive staphylococci (CoPS) cause a range of infections such us bacteraemia, urinary tract infections, pyoderma, abscess and wound infections in both humans and animals [1]. Infections are compounded by the emergence of methicillin-resistant strains that have acquired mecA or mecC imparting resistance to all the beta-lactams with the exception of a few anti-staphylococci cephalosporins [2]. The major methicillin-resistant CoPS that cause clinical infections are methicillin-resistant Staphylococcus aureus (MRSA) in both humans and animals [3], and methicillin-resistant S. pseudintermedius (MRSP) in dogs and cats [4]. MRSA and MRSP isolates are often resistant to multiple classes of critically important antimicrobials (CIAs) including fluoroquinolones and aminoglycosides, thereby limiting therapeutic options to treat these infections. In recent years, studies have demonstrated the emergence and clonal spread of MRSA in companion animals (defined here as dogs, cats and horses), and livestock, with potential for bi-directional transmission of these strains between animals and humans [5][6][7].
In companion animals, distinct MRSA clones appear to colonise specific animal host species. For example, healthcare associated MRSA clone ST22 (EMRSA-15) is most commonly isolated from dogs and cats while community associated MRSA CC 8 (ST8, ST612 and ST254) clones are host-adapted to horses [8]. A recent study by Harrison et al. has also demonstrated that globally disseminated MRSA ST22-IV strains can colonise and cause infection in humans, dogs, and cats without undergoing typical host adaption involving loss or acquisition of antimicrobial resistance and/or prophage genes [9]. These studies demonstrate the need for understanding the ecology and distribution of MRSA clones in companion animals.
Parallel to the emergence of MRSA in companion animals, MRSP has recently emerged in dogs and cats as a cause of skin and soft tissue, post-surgical site and urinary tract infections. Globally, the frequency of MRSP infections in dogs and cats has been increasing and MRSP is now considered to be one of the most important pathogens in small animal medicine [10]. This is attributed to the global spread of MRSP clones and the associated resistance to other CIAs such as fluoroquinolones and aminoglycosides. Unlike MRSA in companion animals, MRSP is not a major zoonotic pathogen and has limited public health impact [10]. However, due to the limited therapeutic options to treat MRSP infections they are now a major animal health issue and require careful monitoring and management [11].
Various studies have demonstrated carriage and zooanthroponotic transmission of MRSA and other multidrug-resistant staphylococci [12] between animals and humans. Consequently, many countries have established surveillance programs to monitor emerging antimicrobial resistance in animals, although companion animals are generally poorly represented in these activities. The frequency and antimicrobial resistance profile of clinical staphylococci in companion animals has been reported in Asia [13,14], Africa [15], North America and Europe [16]. Sweden and Norway are among the few countries that monitor the occurrence of these resistant strains regularly [17,18], enabling more accurate estimation of frequency, trends and antimicrobial resistance profiles to be compared on a yearly basis.
In Australia, several one-off studies have documented the recent emergence of MRSA and MRSP as causes of clinical infections as well as carriage by healthy companion animals [11,[19][20][21]. In addition, carriage of MRSA by Australian veterinarians involved in clinical practice has been well documented with the highest rates of carriage in equine veterinarians (21.4%), mixed-practice veterinarians (11.8%) and those who work exclusively with dogs and cats (4.9%) [7,19,22]. However, the frequency of methicillin resistance amongst isolates from infections in companion animals on an Australia-wide level is currently unknown. Therefore, in the present study, we undertook the first national survey of antimicrobial resistance in CoPS isolated from clinical infections in Australian companion animals. The aims were to define the distribution of CoPS species causing clinical infection in companion animals, the frequency of antimicrobial resistance (particularly methicillin resistance and multidrug resistance) and to examine potential risk factors that may contribute to the occurrence of methicillin-resistant strains amongst the most prevalent species.

Isolate collection and identification
The CoPS isolates were collected during the first national survey of antimicrobial resistance in Australian animal pathogens, which took place over 12 months (January 2013 to January 2014) with the cooperation of all veterinary diagnostic laboratories in each Australian state and territory (n = 22) [23]. Submitting laboratories were instructed to forward coagulase-positive isolates that were considered to be clinically relevant to the presenting condition, as judged by the diagnostic microbiologist. The bacteria were isolated from swabs taken from site of infections or clinical specimens (e.g. urine, biopsies) collected by veterinarians and submitted to veterinary diagnostic laboratories for routine culture and susceptibility testing. All confidential information such as animal name, owner name, address and contract information was removed by the participating veterinary diagnostic laboratories before sending the isolates and clinical information to The University of Adelaide reference laboratory for this study. As a result this study did not require animal ethics approval, as per the Australian National Health and Medical Research Council, Animal Research Ethics code.
Prior to cryopreservation in 20% glycerol broth, isolates were confirmed for purity and haemolysis pattern on Columbia sheep blood agar (SBA; Thermo Fisher Scientific Australia), and identified to genus level using standard phenotypic tests including Gram-stain and the catalase test. A total of 888 isolates were collected in this study, originating from dogs (n = 743, 83.7%), cats, (n = 77, 8.7%) and horses (n = 68, 7.7%). To confirm the identity of staphylococci isolates to species level, all isolates were subjected to MALDI-TOF (Bruker) according to the manufacturer's protocol for bacterial identification.
For S. pseudintermedius, veterinary specific breakpoints were used for AMK, AMC, CLI, ENR, MRB, PRA and TET; human interpretative criteria were used for CHL, CIP, OXA, RIF and SXT and ECOFF criteria as defined by ECOFFinder were used for CVN, FOX, CRE and CEF. For S. aureus, veterinary specific breakpoints were used for AMK, AMC, CEF, CLI, ENR, MRB, PRA and TET; human interpretative criteria were used for FOX, CHL, CIP, OXA, RIF and SXT; ECOFF criteria were used for CVN (defined by ECOFFinder) and CRE (defined by EUCAST). In this study, we used breakpoints for CEF of !0.5 μg/mL instead of !8 μg/mL for S. pseudintermedius as stated in CLSI VET01S in order to correspond with ECOFF criteria and presence of mecA genes in the isolates. Also, for dog and horse isolates, the veterinary specific breakpoint for AMC of !1/0.5 μg/mL was used for isolates from skin and soft tissue infections (SSTIs) and the breakpoint !16/8 μg/mL (non-susceptible) was used for isolates from urinary tract infections (UTIs). For cat isolates, a breakpoint for AMC of !1/0.5 μg/mL was used for both SSTIs and UTIs.

Confirmation of methicillin resistance status
Phenotypic confirmation of methicillin resistance status for putative MRSA and MRSP strains was assessed using resistance to cefoxitin and/or oxacillin, as well as colony appearance on Bril-liance™ Agar MRSA2 (Thermo Fisher Scientific, Australia). Additionally, mecA PCR [31] was undertaken on all S. aureus isolates with cefoxitin MICs !8 μg/mL and all S. pseudintermedius isolates with oxacillin MICs !0.5 μg/mL as recommended by CLSI VET01S [25]. Methicillinresistant staphylococci were reported as resistant to all penicillins, cephems and β-lactams/β-lactamase inhibitor combinations regardless of in vitro test results with those agents [24].

Risk factor analysis
Of the 794 CoPS isolates analysed in this study, a total of 661 (dogs n = 597, 90.3%; cats n = 16, 2.4%; horses n = 35, 5.3%) were accompanied by a detailed clinical history. However, due to low sample size, only S. pseudintermedius isolates from dogs (n = 555) were further interrogated in the risk factors study. S. pseudintermedius that were methicillin-resistant (n = 68) were used as the outcome in this analysis. The variables for potential risk factors were gender (male or female), age group (<2 years, 2-10 years or >10 years), previous antimicrobial treatment (yes/ no), chronic and/or recurrent diseases (yes/ no), and site of infection (ear, urinary tract, skin and soft tissue, surgical site or respiratory tract). Initially, univariate analyses were used to assess the effect of various factors on the frequency of methicillin resistance in S. pseudintermedius isolates from dogs. This was followed by construction of a multivariate logistic regression model to account for the possible effects of confounding and interaction. Age of animal was forced into the multivariate model as a probable confounder and then each explanatory variable was assessed for its significance on the outcome. The most significant explanatory variables were then added to the model and the process repeated (by adding only significant variables) to obtain a main effects model. Two-way interactions between the main effects variables were then explored and retained when significant at P<0.05. Statistical analyses were performed using Stata/MP 14.0 (Stata Corp., College Station, TX, USA).

Distribution of staphylococci species
Of the 888 isolates from companion animals submitted by Australian veterinary diagnostic laboratories for this study, a total of 877 isolates (98.8%) were confirmed to belong to the Staphylococcus genus. The most commonly identified CoPS were S. pseudintermedius (n = 629) and S. aureus (n = 117). Other CoPS identified included S. schleiferi (n = 44), S. intermedius (n = 2) and S. delphini (n = 2). Of the 629 S. pseudintermedius isolates, 97.9% were obtained from dogs and 2.1% from cats. Of the 117 S. aureus isolates, 45.3% were recovered from horses, 40.1% from dogs, and 14.5% from cats. All S. schleiferi isolates originated from dogs while one S. intermedius was isolated from a dog and a cat, respectively. Both S. delphini isolates came from horses. A small number of coagulase-negative staphylococci were also identified, as either Staphylococcus felis (n = 34) isolated from cats, Staphylococcus epidermidis (n = 10) from dogs and cats and Staphylococcus sciuri (n = 10) from horses and dogs. Coagulase-negative isolates were excluded from further analyses.

Phenotypic antimicrobial resistance among S. pseudintermedius
The MIC distribution and frequency of antimicrobial resistance among S. pseudintermedius is shown in Table 2. Among 81 isolates with oxacillin MICs !0.5 μg/mL, a total of 74 isolates (11.8%, dog n = 72, cat n = 2) were classified as methicillin-resistant S. pseudintermedius. The remaining isolates (n = 7) were classified as methicillin-susceptible because of negative results either on the basis of mecA PCR and colony appearance on Brilliance™ MRSA 2 Agar (S1 Table). Among dog isolates (n = 616), resistance to AMC (37.5% of isolates; 45.1% for SSTI, 3.5% for UTI) was most common followed by SXT (37.3%) and TET (22.7%). CLI resistance was observed at a moderate level (12.7%). Similarly, a moderate level of resistance was observed to fluoroquinolones, ranging from 6.5%-8.8% for the four compounds tested in this study (CIP, ENR, MRB, PRA). Resistance to CHL was observed in 5.7% of isolates. A very low number of isolates were resistant to AMK (1.1%; n = 7) and RIF (1%, n = 6).
Among cat isolates (n = 13), the most common resistance found was to AMC (53.8% of isolates) and SXT (30.8%). CHL and CLI resistance was detected in 7.7% isolates. Resistance to AMK, RIF and fluoroquinolones was not detected. Phenotypic antimicrobial resistance among S. aureus The MIC distribution and frequency of antimicrobial resistance among S. aureus is shown in Table 3. Overall, 12.8% of the S. aureus isolates were methicillin-resistant, including six isolates from horses (11.3%), six isolates from dogs (12.8%), and three isolates from cats (17.6%). Among methicillin-susceptible S. aureus, resistance to at least one or more β-lactam antimicrobials was observed in three isolates (2.6%). Among isolates from horses (n = 53), resistance to AMC and TET was high (47.1% and 32.1%, respectively). Resistance to SXT was observed in 13.2% of isolates and resistance to AMK and RIF in 9.4% of isolates. Resistance to CHL and fluoroquinolones was observed at a low level (1.9%). Resistance to CLI was not observed.
Among dog isolates (n = 47), AMC (57.4% of isolates; 59.5% for SSTI, 0% for UTI) and CVN (14.9%) had the highest rates of resistance. Resistance to fluoroquinolones was observed in 8.5% of isolates. A low frequency of resistance (2.1%) was observed for AMK and CHL.

Resistance profiles of S. pseudintermedius isolates
The resistance profiles of the S. pseudintermedius isolates are presented in Table 4. In total, 51.2% of S. pseudintermedius isolates were fully susceptible to eight antimicrobial classes. The proportion of single drug resistance in S. pseudintermedius was 38.1%, with single SXT resistance the most common pattern (18.8%). MDR was observed in 83 isolates (13.2%) including 74 isolates that were regarded as MRSP based on phenotypic characteristics and mecA PCR and nine S. pseudintermedius isolates that were methicillin-susceptible. The most common MDR pattern was resistance to phenicols, lincosamides, fluoroquinolones (FQN), β-lactams,

Resistance profiles of S. aureus isolates
The resistance profiles of S. aureus isolates are shown in Table 5. In total, 68.4% of S. aureus isolates were fully susceptible to eight antimicrobial classes. MDR was detected with a frequency of 12.8%, including six MRSA isolates from horses, six MRSA isolates from dogs and three MRSA isolates from cats. Resistance to TET was the most common pattern observed in horse isolates (17%) and dog isolates (6.4%). Dog and cat isolates were more likely to be resistant to fluoroquinolones (OR 5.4, 95%CI 0.6-252.1), which was also always associated with methicillin resistance, compared to horse isolates. Horse isolates were more likely to be amikacin-resistant (OR 6.5, 95%CI 0.7-315.2) compared to dog and cat isolates. All rifampicin-resistant S. aureus isolates from horses (n = 5) were methicillin-resistant.

Risk factors for MRSP in dogs
In univariate analysis, there was no significant difference in the proportion of MRSP isolates between female versus male dogs; chronic versus non-chronically diseased dogs; or the various age groups ( Table 6). Site of infection and prior antimicrobial treatment were significantly associated with MRSP isolation and were retained in the multivariate model. In multivariate analysis, after controlling for the confounding effect of age, isolates from particular infection   (Table 7).

Discussion
This is the first comprehensive study describing the distribution of antimicrobial susceptibility profiles in CoPS isolated from clinical infections in companion animals in Australia. This study generated three major findings: 1) The frequency of MRSP and MRSA isolation from clinical infections in companion animals in Australia was estimated as moderate (11.8% and 12.8% of total isolates for each species, respectively); 2) Resistance to critically important antimicrobials used in human medicine (fluoroquinolones, amikacin) remains very low to low among Australian companion animal CoPS; and 3) Prior antimicrobial treatment was identified as a significant risk factor for isolation of MRSP from dogs with surgical site, skin and soft tissue infections. MRSP infections are increasingly reported in veterinary practice, spreading among companion animals and to a lesser extent among veterinarians [10]. The increased frequency of MRSP that are MDR poses a serious concern for biosecurity and infection control in veterinary practices, due to limited therapeutic options and the ease of transmission between animals. In parallel to other noteworthy studies from Australia [11], MRSP isolates were resistant to more antimicrobial classes than MRSA isolates, exemplified by the high proportion of MRSP isolates showing resistance to more than six antimicrobial classes (n = 29, 4.8%), while this level of multidrug resistance was not identified in any S. aureus isolates. Compared to similar surveys in other countries conducted over the same time period, the frequency of methicillin-resistant strains among canine S. pseudintermedius in Australia (11.8%) was significantly higher (P<0.0001) than that reported in Sweden (0.4%) [17] and Norway (0.5%) [18]. However, resistance to clindamycin in S. pseudintermedius in Australia (12.7%) was significantly lower (P = 0.0001) than in Sweden (21.6%). While a high level of amikacin resistance in S. pseudintermedius isolates has been demonstrated in some studies [32], we found that only a very low proportion of companion animal S. pseudintermedius isolates from Australia were resistant to this critically important human drug (n = 7; 1.1%). It is therefore recommended that use of amikacin in veterinary medicine continues to be reserved for MDR infections identified on the basis of culture and susceptibility testing when no other drug class is available [33].
In the only other comparable study conducted in 2006 in two regions of Australia, involving both clinical and non-clinical (i.e. carriage) of Staphylococcus spp. isolates from dogs and cats (n = 331), the frequency of methicillin-resistant (mecA) and β-lactam-resistant (blaZ) strains was only 3% and 6.9%, respectively [20,34]. Although methodologies for sampling, testing and data interpretation were somewhat different to this study, it might indicate that methicillin resistance amongst Australian companion animal staphylococci has substantially increased in less than a decade.
In a recent Australian study, colonisation of veterinarians by MRSA was dominated by strains belonging to CC8 MRSA (ST8-IV [2B], spa t064; and ST612-IV [2B], spa variable). These were strongly associated with equine practice veterinarians and were often resistant to rifampicin and gentamicin [19]. MRSA CC8 (ST8 and ST612) is the most commonly identified clone among both Australian veterinarians and clinical equine samples [35]. Similarly, in the present study, a high proportion of MRSA isolates from cases of infection in Australian horses were also resistant to rifampicin (9.4%) but rarely resistant to fluoroquinolones. Rifampicin is almost exclusively used in equine practice, where it is combined with a macrolide for the oral treatment of Rhodococcus equi infections in foals [35,36]. The equine MRSA isolates identified in the present study were sensitive to range of additional antimicrobial classes including chloramphenicol and fluoroquinolones, demonstrating that additional therapeutic options were still available for treating MRSA infections in horses. The significant association between methicillin resistance and fluoroquinolone resistance in Staphylococcus spp. isolates from dogs and cats in this study reflects the observation that a high proportion of Australian MRSP isolates may belong to internationally disseminated fluoroquinolone-resistant clones such as ST71 and ST45 [11,37], whereas MRSA isolates are likely to belong to ST22-IV [2B], spa variable, commonly found in small animal practice veterinarians in Australia [19,38] and community-acquired infections [39]. A comparative genomics study is currently underway to determine genetic similarity of methicillin-resistant isolates in this study. Comparative genomics represents the most rapid, cost effective and accurate technique for molecular typing including determination of sequence type.
The most important finding from the risk factors study was that particular sites are associated with the risk of a S. pseudintermedius infection being methicillin-resistant. In agreement with the present study, other studies have also found that isolates from surgical site infections were at higher risk of being resistant to methicillin when compared to other sites [10,40]. In parallel to the work here, animals that were hospitalised, visited veterinary clinics frequently or had previous antimicrobial treatment were at higher risk for MRSP infections [41]. Compared to studies from 2006, [20,34] it appears that MRSP infections are becoming increasingly common in veterinary companion animal practice in Australia. The results strongly reinforce the need for veterinarians to place a high priority on implementing infection control procedures, biosecurity and antimicrobial stewardship such as those recommended by the Australian Veterinary Association [42]. Understanding potential factors that lead to emerging resistance may aid in the development of strategies that could curtail the ongoing spread of MRSP within veterinary hospitals.
This study has some limitations. Inclusion of isolates was performed at the convenience and discretion of the animal owners (who would be expected to pay for tests at the primary laboratory), the consulting veterinarian (who may or may not favour sensitivity testing), and the primary laboratory (who may or may not be interested in the study). Consequently, the resulting size and direction of bias in estimates of resistance frequency is difficult to define. Further, the small sample size of methicillin-resistant feline isolates led to wide confidence intervals, limiting our ability to draw statistically significant conclusions on feline isolates. Future studies should therefore focus on achieving a sufficiently large collection of isolates from cats to increase the accuracy of these estimates. Despite these shortfalls, we are unaware of any collection of isolates that is as representative of the Australian population of companion animals, both in terms of size and geographic diversity. Certainly the data presented here surpasses what is currently available elsewhere in the literature [20,43] and is therefore a useful basis for reviewing prescribing practices for staphylococcal infections in companion animals both in Australia, and more broadly.

Conclusions
This study shows that antimicrobial resistance is commonly present in the coagulase-positive staphylococci cultured from animal infections in companion animals in Australia. Of greatest concern is the occurrence of moderate levels of MRSP and MRSA, some of which are also resistant to fluoroquinolones. The data provides important baseline measurements for future surveillance and international benchmarking. A strong association of MRSP with surgical site infections in dogs suggests that there could be shortfalls in infection control in animal hospitals. Periodically repeated surveys of this type are crucial for understanding the trends in emergence and dissemination of antimicrobial resistance in companion animals.
Supporting information S1 Table. Determination of methicillin resistance in S. pseudintermedius and S. aureus isolates from dogs, cats and horses in Australia based on phenotypic characteristic and mecA PCR. Table. Percentage of resistance to penicillins, cephems and β-lactams/β-lactamase inhibitor combinations before and after confirmation of methicillin resistance status. (DOCX)