Surveillance of antimicrobial resistance of maltose negative Staphylococcus aureus in South African dairy herds

Background The discovery of antimicrobials in the 1930s was one of the greatest achievements in medicine. However, bacterial resistance to antimicrobials was already observed in the 1940s and has been reported since then in both human and veterinary medicine, including in dairy cows. Many years of monitoring milk samples in South Africa, has led to the identification of a new strain of Staphylococcus aureus (S. aureus), which is maltose negative and appears to be an emerging pathogen. In this study the differences in susceptibility to antimicrobials of this strain were evaluated over time, over different seasons, in different provinces, and according to somatic cell count (SCC) categories. Results A data set of 271 maltose negative S. aureus isolates, cultured from milk samples from 117 herds out of the estimated 2000 commercial dairy herds in South Africa between 2010 and 2017, was studied using the disk diffusion method. This analysis was done using the Clinical Laboratory Standards Institute (CLSI) breakpoints in order to compare using both the previous (Intermediate category grouped with Resistant) and current definitions, (Intermediate category grouped with Susceptible). The results of the analysis between the previous and the current definitions differed for tylosin, cefalonium, oxy-tetracycline and cloxacillin. Neither the analysis using the previous nor the current systems showed an effect of province for the maltose negative S. aureus. This was in contrast to the results for maltose positive S. aureus where differences between provinces were shown in a previous study, with the lowest prevalence of resistance shown in KwaZulu-Natal during spring. For the susceptibility testing of 57 maltose negative and 57 maltose positive S. aureus isolates from 38 farms, from KwaZulu Natal, Eastern Cape and Western Cape. The minimum inhibitory concentration (MIC) results for the maltose negative S. aureus isolates confirmed the results of the disk diffusion method. differed general, in their antimicrobial resistance patterns over time, in comparison to maltose-positive S. aureus strains. MIC testing also indicated more multidrug -resistant isolates seen maltose negative aureus than in the maltose positive


Background
The genus Staphylococcus consists of a variety of opportunistic pathogens of variable relevance in veterinary medicine. The most clinically relevant staphylococci in veterinary medicine are the coagulase positive Staphylococcus aureus and members of the S. intermedius group (SIG) [29], particularly Staphylococcus pseudintermedius (S. pseudintermedius). A noted property of staphylococci is their ability to become resistant to antimicrobials. Methicillin resistance is of particular relevance, because it is conferred by a presence of the mecA gene, which encodes for the production of an altered penicillin binding protein (PBP) (PBP2a or PBP2') that has a low affinity for all beta-lactam antimicrobials (penicillins, cephalosporins, carbapenems) [17]. Methicillin-resistant S. aureus (MRSA) is recognised as a significant problem in human medicine and it is among the most important infections in hospitalized individuals and in people in general [16].
In veterinary medicine S. aureus (maltose positive) is the biggest problem in the dairy industry in South Africa and globally [24]. The infected udder is considered the primary reservoir of S. aureus and the organism is believed to be transmitted during milking. Despite this, a proportion of heifers which are already infected with S. aureus, enter the milking herd [23]. This suggests routes of transmission in addition to the milking equipment and the milking parlour. A good understanding of S. aureus reservoirs and transmission is essential for the effective control of the organism in a herd. The probability of treatment resulting in cure of S. aureus (maltose positive) infection is calculated by taking the following factors into account: parity; the stage of lactation; the SCC level; the specific teat position on the udder; the number of quarters infected; and the duration of treatment required [32].
The Milk Laboratory at the Faculty of Veterinary Science at the University of Pretoria has provided an extensive dairy cow udder monitoring programme in South Africa. Since 2005, an increasing number of coagulase positive, maltose negative staphylococci have been isolated, that were confirmed as maltose negative S. aureus by molecular methods. These organisms were first identified from a dairy cow in a single South African dairy herd with a somatic cell count of less than 100 000 cell/ml of milk.
As early as three years later these organisms were isolated from numerous other dairy herds in South Africa, albeit with effective susceptibility to antimicrobials that were tested routinely and with a low SCC. However in more recent years 2016/2017/2018, coagulase positive and maltose negative staphylococci were isolated showing resistance (MRSA, cefoxitin disc) where some of the milk samples started to show higher SCC (> 400 000cells/ml milk) than initially shown (personal experience). In addition to the previous study which attempted to characterise this emerging pathogen, a further evaluation was carried out of the resistance trends evident in historic disc diffusion susceptibility data and more recent MIC data.
The main objective of this study was to investigate the retrospective (disc diffusion) antimicrobial surveillance data of maltose negative S. aureus in different provinces, seasons and SCC categories, and to determine any differences between the use of the previous versus the most recent CLSI classification systems. An additional objective was to compare the MIC results of maltose positive and maltose negative S. aureus (emerging pathogen) and to discuss the practical implications thereof.

Results
The first part of this study was the retrospective data analysis (disc diffusion) which was done only on maltose negative S. aureus, as a previous study had been done with a similar analysis of maltose positive S. aureus that combined the resistant and intermediate results and reported susceptible results separately [14]. The eight antimicrobials that were used in this retrospective study were the commonly used antimicrobials that are available as intramammary remedies in South Africa. The original classification [6; 7] and the more recent classification (intermediate grouped with susceptible) [8; 9] of antimicrobial resistance showed similar trends. These trends of resistance to ampicillin, cephalexin, cefalonium, cloxacillin, oxy-tetracycline and penicillin peaked (at highest) in 2011, and for tylosin in 2013, and then decreased over time. However, in 2014 there was a slight increase in antimicrobial resistance seen for cloxacillin and in 2016 for ampicillin, cephalexin, cefalonium, penicillin and tylosin (Figures 1, 2 , 3, 4, 5 & 6). The analysis of antimicrobial resistance for cloxacillin showed significant differences between SCC categories only according to the GLMM analysis (p<0.05). Cefoxitin, cloxacillin and cefuroxime showed no significant differences for years and provinces ( Table 1). The analysis of antimicrobial resistance of maltose negative S. aureus for oxy-tetracycline, cephalexin, ampicillin, tylosin, cefalonium and penicillin G showed significant differences according to the GLMM (p<0.05) over time (years). Similar apparent trends (but not significant) were shown for clindamycin which was used from 2009 to 2012 and an apparent increase in percentage of antimicrobial resistance and (also not significant) for cefoxitin which was used from 2014 to 2017 and showed a decreased antimicrobial resistance over time.
These data were re-evaluated according to the more recent classification system [8; 9] that grouped results from susceptible and intermediate data together, so that the resistant data were reported separately. Some differences from the previous analysis were apparent as a detailed examination of Tables 1 and 2 will show. Tylosin, cloxacillin, oxy-tetracycline and cefalonium showed differences in the effects of year, season, province and SCC categories on antimicrobial resistance as seen between the original system of classification (Table 1) compared to the more recent system of classification (Table 2) using the CLSI guidelines. According to the new guidelines, tylosin, oxy-tetracycline and cefalonium showed no significant effects at all on antimicrobial resistance from the factors measured (Table 2), whereas they had showed some effects of year, season and SCC categories when using the previous classification (Table 1). Cloxacillin showed an effect of SCC category only when using the previous classification system for analysis (Table 1), but there appeared to be possibly a slight effect of season as well as when using the more recent classification (Table 2). There was no effect of province on antimicrobial resistance when using both the previous system (Table 1) and the more recent system (Table 2) of the CLSI guidelines. Ampicillin, penicillin G, cefoxitin and cefuroxime showed similar effects of year, province, season and SCC categories for both the previous system (Table 1) and the more recent system (Table 2) CLSI guidelines. For ampicillin, comparing the results from the previous system of analysis, seasonal effects showed differences only for autumn compared to spring (Table 1). However, by using the more recent system, the results for ampicillin appeared to show an effect of both summer and autumn (Table 2) with high resistance at those times, and this could be compared to spring where there was low resistance.
This comparison between the analysis of the effects of year, season, province and SCC categories of antimicrobial resistance of maltose negative S. aureus shows that the new breakpoints (CLSI) as well as the grouping of intermediate readings with susceptible readings instead of with resistant readings, as was done in the past, does indeed make a difference to the results (Tables 1 & 2). The logic behind the change of this grouping, was that in the past, intermediate readings were grouped with resistant readings to create more strictly defined categories.
Antimicrobial products shown in Tables 3 and 4 are all the products from the PM 32 panel (Beckman Coulter) that showed resistance to any of the isolates tested. The 57 maltose positive and the 57 maltose negative S. aureus isolates were all resistant to ampicillin. Out of the total of 114 isolates overall, only 37 were resistant to more than one product, 30 maltose negative S. aureus and seven maltose positive S. aureus (Table 4). There were a total of 25 multidrug resistant (MDR) isolates (isolates resistant to an antimicrobial from three or more antimicrobial classes), 3 maltose positive and 22 maltose negative S. aureus (Table 4).

Discussion
This study on the maltose negative S. aureus showed no significant differences of antimicrobial resistance between the provinces, and only a limited significant difference related to seasons and SCC categories, whereas there were no significant interactions between any of the variables considered (Tables 1 & 2). The relationship of the occurrence of SCC category to antimicrobial resistance was in agreement with a study in Denmark [5], which also found high SCC to correspond with low antimicrobial resistance and low SCC to correspond with high antimicrobial resistance. This could be due to the SCC being more of an indicator of irritation and severity of the infection rather than an indicator of antimicrobial resistance of the organism.
In contrast, the study on maltose positive S. aureus showed that when considering the provinces, the lowest prevalence of antimicrobial resistance to the majority of the categories of antimicrobials that were tested was present in KwaZulu-Natal during spring, except for cephalosporins which had the lowest levels of prevalence of bacterial resistance in Gauteng during winter [15]. Although, there were great differences in the numbers of herds and samples between the provinces, these differences were taken into account in the model, during the analysis.
Resistance patterns of the maltose positive S. aureus to the eight antimicrobials varied in the different seasons and provinces, possibly because of the different weather conditions, as well as the action and spectrum of antimicrobials [15]. There was one specific strain of maltose negative S. aureus identified, originating in and found mostly in KwaZulu Natal, but also now present in all nine provinces of South Africa, albeit in small numbers.
The antimicrobial resistance trends (Figures 1 to 6) were in agreement with those shown for the same antimicrobials with coagulase negative staphylococci [27], but in contrast to the trends shown for the maltose positive S. aureus over time [14]. The study on the maltose positive S. aureus showed a general increase in resistance over time except for the 20 well managed herds (part of the pro-active udder health programme), which showed a decrease in resistance over time [14]. However, due to  (Tables 3 & 4), allow for informed treatment choices to be made without the need to wait for any specific antimicrobial sensitivity test results. However, the results of antimicrobial sensitivity tests (disc diffusion or MIC), are just an indication that the particular organism has the ability to be killed by a particular antimicrobial (invitro). In the udder, the situation may be very different from tests in vitro. This is because the site of the infection may be very difficult to reach via very small arteries and lactiferous ducts and also due to udder pharmacodynamics (very few products are successful in a water and fat environment). As a result of this, the treatment success for mastitis in is not very high (27%) as has been determined in a large study done in five European countries; France, Hungary, Italy, the Netherlands, and the United Kingdom [33] and can lead to the development of antimicrobial resistance by mastitis-causing organisms. Treatment success against S. aureus tends to be better in the dry period but still not ideal [3], therefore the pro-active udder health programme should be applied. This is why as far as mastitis is concerned, the focus should be on the prevention and monitoring through the pro-active udder health programme [25] rather than on the actual treatment.
The second part of this study concerned the MIC test on 57 maltose negative and 57 maltose positive S. aureus isolates. The MIC results of antimicrobial resistance (Table 3)  Antimicrobials contribute to animal care in four ways: to treat animals diagnosed with an illness; to control spread of illness within a herd or flock; to prevent illness in healthy animals when exposure is likely; and to ensure healthy growth by maintaining the right balance of bacteria for improved nutrient utilization (antimicrobials approved for animal use only) [10]. Antimicrobials that are used in animals are important for global food security. Estimates have been made that nearly one billion people in the world do not get enough to eat each day and nearly three billion are trying to diversify their diet to include more meat, milk and eggs [10]. Veterinary medicines including the antimicrobials provide a valuable tool to help veterinarians and producers to deliver healthy animals to meet the growing need for safe, nutritious, affordable food, while making the most of limited natural resources in a sustainable manner. As the world population is expected to grow to nine billion by 2050, tools such as antimicrobials which keep animals healthy will be essential to meet the increasing demand for food [10].
Antimicrobials that could be considered for routine testing by veterinary diagnostic laboratories, may be divided into four groups [11]: Group A: Antimicrobials with specific interpretive criteria for veterinary medicine; Group B: CLSI approved interpretive criteria for human medicine; Group C: No veterinary species-specific or human-specific interpretive criteria; Group D: Supplemental to be tested selectively.
Ampicillin, oxacillin, erythromycin, penicillin and tetracycline are the corresponding antimicrobials from the panel tested which are approved for the control of bovine mastitis specifically.
The distribution of the MIC test results of the antimicrobials tested are summarized in Tables 3 and 4 for maltose positive and maltose negative S. aureus isolates respectively. One maltose positive and 12 maltose negative S. aureus isolates were resistant to oxacillin (Table 4). For erythromycin one maltose positive and nine maltose negative S. aureus isolates proved to be resistant, respectively.
The MICs of clindamycin were two dilution steps lower ( Table 4) than those of azithromycin, cefotaxime, erythromycin, gentamycin, linezolid, teicoplanin, tetracycline, tobramycin, trimethoprim / sulphamethoxazole and vancomycin for maltose negative S. aureus isolates (Table 3). However these patterns differed for the maltose positive S. aureus isolates (Table 3). In general, the resistance rates of the maltose positive S. aureus obtained in this study, corresponded well to those reported in other studies [28; 35].
The MIC 50 represents the MIC value at which ≥50% of the isolates in a test population are inhibited, and it is equivalent to the median MIC value. The MIC 90 represents the MIC value at which >90% of the isolates in the test population are inhibited [30]. The MIC breakpoints (chosen concentration  Table 3). The maltose negative S. aureus was more resistant to the amoxicillin clavulanic acid combination (used in human medicine), ampicillin and cefuroxime at MIC 90, and for clindamycin at MIC 50 (Table 3). However, oxacillin was more resistant for maltose positive S. aureus for MIC 90 and MIC 50 and clindamycin for MIC 90 (Table 3). Infrequently found resistance patterns were found in 17 of the 57 maltose negative S. aureus isolates which were resistant to vancomycin and one maltose positive and eight maltose negative S. aureus isolates which were resistant to oxacillin [11].
Overall in this study there were more multi-drug resistant maltose negative S. aureus than maltose positive S. aureus isolates and the same general interpretation applied for isolates resistant to two or more antimicrobials of varying combinations in general (Table 4).
There have been many studies in both animal and human medicine that have identified multidrug resistant and pan-drug resistant S. aureus isolates [12]. However, most of these studies were done on traditionally identified coagulase positive, maltose positive S. aureus, since the maltose negative coagulase positive isolates were thought at that time to have been part of the S. intermedius group of isolates. Multi-drug resistant S. intermedius and S. pseudintermedius isolates have been described in dogs, cats and horses, mainly from skin infections [20; 36]. Although a coagulase positive, maltose negative S. aureus strain was subsequently isolated from bovine mastitis [13], there appear to be no antimicrobial susceptibility profiles of this organism yet. The maltose negative S. aureus in this study was originally phenotypically identified as S. pseudintermedius [11; 29], but further MALDI-TOF MS and 16S r RNA sequence analysis on these isolates from dairy cattle, showed that these were in fact a strain of S. aureus (Karzis et al. submitted JDS 2019). Therefore it would make sense that these maltose negative S. aureus isolates in this study, have reacted in a way more like the classical S.
pseudintermedius isolates in other studies, identified from other species [20; 36]. One of the limitations of this study is the limited sample numbers of maltose negative S. aureus available to be used for this trial.
Human nasal S. aureus colonization has previously been reported as being a risk factor for pig farming [4] and S. aureus strains from pig farmers were found to be those present in pigs but had not been found in non-farmers [2]. However, little attention was paid to MRSA in pigs until 'unexpected' MRSA infection and colonization were identified in people that had been in contact with pigs in the Netherlands [34]. In the light of this research in pigs, it is possible that in a similar way of transmission of some strains of these resistant S. aureus (predominantly maltose negative strains) isolated from dairy cattle in South Africa could be from people. Previous studies done in KwaZulu Natal [31], have indicated zooanthroponosis ("reverse zoonosis") of S. aureus in South Africa, with one of the strains identified as the same maltose negative S. aureus strain. These maltose negative strains of S. aureus seem to have completely different antimicrobial resistance trends and severity of antimicrobial resistance, when compared to the traditionally identified maltose positive S. aureus and thus need to be treated differently in practice.
Resistance to antimicrobial agents, such as doxycycline and trimethoprim-sulfamethoxazole is very uncommon in S. aureus. In this study the MIC 90 of trimethoprim-sulfamethoxazole was the same for both the maltose negative and the maltose positive S. aureus. The reason for these uncommon resistance profiles of maltose negative S. aureus isolates, which have also shown to be resistant to antimicrobials that are only used in human medicine, (e.g. carbapenems like imipenem and ertapenem) which had 10 resistant isolates each (Table 4), remains questionable. These are antimicrobials which are not used at all in animal medicine. Zooanthroponosis seems to be a strong possibility, because these isolates have been shown to be present on the skin of humans that come into close contact with dairy cattle. This would be similar to the findings of the studies with the dogs in Brazil [19] and pigs in Germany [2] and the Netherlands [35] respectively. Further studies need to be done to explore the origin of such resistant isolates of maltose negative S. aureus. Future work is also necessary to determine the resistance genes present in resistant maltose negative S. aureus strains.
Similarly to this study, a study done in dairy cattle in Tennessee also found that there was a variation of prevalence of antimicrobial resistance of S. aureus within and among farms over time, with an increasing trend in tetracycline resistance [1]. This was also in accordance with the report which showed that the predominant antimicrobial groups used in animal health management in South Africa from 2014 to 2015, were the growth promoters (animal-use-only antimicrobials) (62%) followed by tetracyclines (17%) and macrolides (11%) [21]. In this study five isolates were resistant to tetracyclines and ten isolates were resistant to macrolides (represented by erythromycin) and only one of each of these isolates were maltose positive, whereas the rest were maltose negative S. aureus ( Table 4). Antimicrobial resistance trends over time also differed between maltose negative and maltose positive S. aureus. Antimicrobial resistance of maltose negative S. aureus showed no significant differences between provinces (using both the previous and also the more recent interpretation) and there were only limited differences between seasons for ampicillin, penicillin, tylosin and cefalonium (using the more recent interpretation system). For cloxacillin and cefalonium specifically, high SCC corresponded with low antimicrobial resistance and were significantly different from low SCC which to corresponded with high antimicrobial resistance. The MIC results of antimicrobial resistance for maltose negative S. aureus confirmed the results of the disc diffusion method. The results of the MIC method also showed more resistance in general for maltose negative than for the maltose positive S. aureus isolates to most of the antimicrobials that were used. The MIC breakpoints were susceptible at MIC 50 and MIC 90 for both maltose negative and maltose positive S. aureus, with the exception of maltose negative S. aureus at MIC 90.

Data collection and bacterial identification
In an initial retrospective study 271 isolates from 117 dairy herds out of approximately 2000 commercial dairies in South Africa [18] were analysed from 2010 to 2017. These herds were located in all nine provinces of South Africa but with greatly varying numbers in the different provinces, namely: Gauteng (n= 8), KwaZulu Natal (n= 170), Free State (n= 4), Eastern Cape (n= 56), Western Cape (n= 27), Northern Cape (n= 1), North West (n= 1), Limpopo (North) (n= 1) and Mpumalanga (n= 3). All herds from which these isolates were collected participated in the pro-active udder health programme (routine testing of microbiology and cytology) at the milk laboratory [14]. The milk samples were collected in an aseptic manner according to a standard operating procedure [22]. These isolates were collected and cultured according to the method recommended by the National Mastitis Council [22] and somatic cell counts were performed using a Fossomatic 5000 and Fossomatic FC (Rhine Ruhr). Isolates to be tested for routine antimicrobial susceptibility were all selected from milk samples with a somatic cell count (SCC) (Fossomatic 5000 and Fossomatic FC, Rhine Ruhr) of more than 400 000 cells/ml [23], when applicable, in order to include cases of subclinical mastitis. This was the general rule for routine antimicrobial susceptibility testing. However, when a maltose negative S. aureus was isolated from a herd, antimicrobial sensitivity testing was done on that organism regardless of SCC. Phenotypic differentiation of staphylococci was initially identified based on colony morphology, pigmentation, haemolysis, catalase, staphylase, maltose and potassium hydroxide tests that were used [26]. A positive maltose agar reaction confirmed S. aureus and a negative maltose reaction confirmed an organism potentially from the S. intermedius group [11; 29].

Disc diffusion method
The disc diffusion method [4] was used to determine the antimicrobial susceptibility of the routine diagnostic samples that were used for the retrospective data analysis. The initial results were based on the diameter of the inhibition zones and were classified as sensitive, intermediate or resistant Nine antimicrobials used in intramammary treatment (dry and lactating remedies) that were available for use in South Africa were tested. These were the penicillins (ampicillin 10 μg, cloxacillin 5 μg, penicillin G 10 IU), cephalosporins (cephalexin 30 μg, cefuroxime 30 μg), lincosamides (clindamycin 10 μg), tetracyclines (oxy-tetracycline 30 μg) and macrolides (tylosin 30 μg) and (cefoxitin 30 μg).
The lactating cow numbers of the herds in this study varied from approximately 30 (smallest herd) to 1 700 cows, (largest herd) [18].

MIC method
In addition, MIC tests were performed on 57 coagulase positive, maltose negative S. aureus isolates

Statistical Analysis
The retrospective data were originally analysed according to the previous conventional system [6; 7], as described above. However, after the introduction of the new CLSI [8; 9] recommendations, this analysis was repeated using the recently introduced system [8 ; 9], and the two sets of results were compared.

cells per ml milk).
The Chi-square test was used to check for the existence of any effect of year, province, season or SCC category on the response variable, the resistance to the different antimicrobials. This Chi-square test also allowed classification of the categories of each variable in order to introduce the one with the lowest level of resistance as the reference category in the following GLMM (general linear mixed model) analysis. Apparent relationships between season, province and SCC category on the prevalence of antimicrobial resistance of all S. aureus isolates were tested in a multivariate model, a ('glmer' function within 'lme4' package of the R software © version 3.3.3), with a 95% confidence interval using a logit link-function. This model allowed the analysis to take into account the random effect from the different herds. This type of general linear mixed model (GLMM) takes into account this random effect when comparing the different provinces, the different seasons and the SCC category as well as the interactions. Under the "goal of parsimony", a stepwise approach based on the smallest Akaike Information Criterion, was used to select the best model. The study was a retrospective analysis approved by the University of Pretoria Ethics Committee (reference number V062/14). The laboratory that supplied the data provided written consent from owners for the data to be used for research purposes and this was approved by the ethics committee.

Consent for publication
The study does not involve human subjects and therefore no consent was required. However, the laboratory that supplied the study data provided consent for study results to be published.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This research was partly funded by the National Research Foundation. The funding body had no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.      A  S  S  S  S  S*  STA+  2  1  S  R  S  S  S  S  S  S  S  S  S  S  S  S  R  S  STA-2  1  S  R  S  S  S  S  S  S  S  S  S  S  S  S  S  S  STA-2      Trends of antimicrobial resistance to ampicillin over time of maltose negative S. aureus.
Resistance mean is the proportion of isolates per year; this figure used the disc diffusion, retrospective data, according to CLSI [8; 9], with intermediate responses grouped with susceptible responses (the more recent system).

Figure 5
Trends of antimicrobial resistance to cephalexin over time of maltose negative S. aureus Resistance mean is the proportion of isolates per year; this figure used the disc diffusion, retrospective data, according to CLSI [8; 9], with intermediate responses grouped with susceptible responses (the more recent system).

Figure 6
Trends of antimicrobial resistance to penicillin G over time of maltose negative S. aureus.
Resistance mean is the proportion of isolates per year; this figure used the disc diffusion, retrospective data, according to CLSI [8; 9], with intermediate responses grouped with susceptible responses (the more recent system).