Uterine Microbiology and Antimicrobial Susceptibility in Isolated Bacteria from Mares with Fertility Problems

Uterine microbiology and antimicrobial susceptibility was investigated in 239 mares with fertility problems in a prospective study in Sweden. Uterine swab samples were collected with double guarded swabs and transported overnight before being cultured. The Minimum Inhibitory Concentrations (MIC) was determined for a panel of antimicrobials. From 152 of the 239 mares at least one bacterial species was isolated, most frequently E. coli (104 isolates), β-haemolytic streptococci (31) and fungi (16). β-haemolytic streptococci were more frequently (p < 0.01) associated with clinical endometritis than with repeat breeding. The opposite was true for E. coli (p < 0.01). Among β-haemolytic streptococcal isolates some resistance was noted for 4 of 11 tested antibiotics, however, all isolates were susceptible to the widely used penicillin G. Among E. coli isolates enrofloxacin was the only of the 10 tested antibiotics for which no resistance was noted. Resistance was most commonly noted to cephalothin (39% of the isolates), streptomycin (22%), trimethoprim/sulphamethoxazole (15%) and ampicillin (11%). In conclusion, we show that both E. coli and β-haemolytic streptococci are frequently associated with fertility problems in mares and that antimicrobial resistance is a common feature of E. coli but also recognised for β-haemolytic streptococcal uterine isolates.


Introduction
Uterine infections have long been recognised as one of the major causes of reduced fertility in the mare (Asbury 1986). These infections are most often caused by opportunistic micro-organisms and a variety of species have been isolated (Shin et al. 1979, Ricketts et al. 1993. The uterine infections often cause endometritis. Antibiotics are one component often used in the treatment of endometritis (Perkin 1999). For clinicians there is a need of rapid microbiological diagnosis so that adequate treatment of the infection can be performed while the mare is still in oestrus (Ricketts & Mackintosh 1987, Ricketts 1989. Therefore some mares are treated with antibiotics without a preceding microbiological investigation, whereas sometimes bacteriological cultivation is performed by the clinicians themselves. If the treatment is performed without a microbiological diagnosis, the choice of antibiotic is often based on data from earlier studies, e.g. Shin et al. (1979) and Ricketts et al. (1993). However, the bacterial species isolated, as well as their susceptibility to antibiotics, may vary over time as well as from one population of horses to another (Sternberg 1999). The variation may be attributable to differences in antibiotic treatment policies, studfarm management, breed and clinical history of the sampled mares as well as microbiological culture routines. The present survey of uterine microbiology and antimicrobial susceptibility in mares selected for having fertility problems was conducted in Sweden where artificial insemination is commonly used. Also the policies for clinical use of antibiotics are regarded to be strict in Sweden (Franklin 1999, SVARM 2001. The aim of the present study was to determine the most common bacterial species in uterine samples from Swedish mares with fertility problems and the antimicrobial susceptibility of isolated bacteria. Such data should serve as a basis for updated recommendations on how to treat uterine infections in the mare.

Sampling
Clinicians representing different types of stud farms, geographically located all over Sweden, were invited to send uterine samples from mares in oestrus to the National Veterinary Institute (SVA) for culture free of charge. During the spring and summer 1996 and 1997 swabs from 239 mares were submitted from 36 different clinicians.

Mares
For all 239 mares included in the study, at least one fertility problem was noted. From the preceding season 89 barren mares and 33 abortions/resorbtions were recorded. During the current season 121 repeat breeding mares were recorded. Repeat breeding was defined as starting a new oestrus cycle after artificial insemination (AI) or being bred by a stallion once or repeatedly during oestrus in at least one oestrus cycle (with a normal or changed length of the luteal phase). Clinical signs of endometritis during the current season were noted in 89 mares. The designation clinical signs of endometritis in this study included at least one of the following criteria: vulvular discharge or fluid in the uterine lumen during the luteal phase, the latter diagnosed with ultrasonography, inflammatory cells on a cytological smear sample or significant bacterial growth. Reproductive problems during the current season may be combined with barreness or resorbtion/abortion during the preceding season. Repeated breeding was often combined with some of the other reproductive disorders. The average age of the mares was 12.2 years. For 32 mares the age was not recorded, 32 mares were from 3-7 years, 85 from 8 to 13 years and 90 from 14 to 24 years old. The dominating breeds of the included mares were Swedish Warmblood (80), Standardbred Trotter (75), North-Swedish Trotter (39) and Thoroughbred (25). Twenty mares were either of other breeds or their breed was not recorded. Since many clinicians were involved in the present study, it was important to establish a simple, straightforward sampling protocol in order to achieve good quality and reliable data. Hence, we limited the study to bacteriological sampling. For these samples, the external genitalia were carefully washed with soap and water and thereafter dried with paper. In order to minimise contamination of the sample by bacteria from the vagina and perineum the sampling was performed using a gloved hand in the vagina and double-guarded, occluded swabs enabling sampling from the uterus solely (Equi-Vet, Kruuse, Marslev, Denmark). The uterine culture swabs were transported in Amies' modified media with charcoal (SVA, Uppsala, Sweden) (Amies 1967, Engvall 1985 at ambient temperature and cultured within 24 h. This medium has been widely used in Sweden as an all-purpose transport medium for equine gynaecological swabs.

Bacteriology
All samples were cultured on 5% horse blood agar (SVA) and lactose bromocresol purple agar (SVA). The samples were inoculated on the agar plates and diluted by an inoculation loop obtaining 3 levels of dilution on the agar plate. Each bacteriological culture was inspected and bacterial growth was registered after 24 and 48 h incubation at 37°C . Growth of Pseudomonas (P.) aeruginosa, Klebsiella (K.) pneumoniae, haemolytic Escherichia (E.) coli and ß-haemolytic streptococci was always considered to be of significance (Shin et al. 1979, Ricketts et al. 1993, Langoni et al. 1997. Other bacterial isolates were typed and considered as significant if growth was in pure culture or dominating on the agar plate. From the same sample, 2 bacterial species might be isolated and typed. Bacterial growth was evaluated on horse blood agar plates according to the following guidelines: abundant, >100 CFU (colony forming units)/plate; moderate, 21-100 CFU/plate; sparse, 10-20 CFU/plate; insignificant, <10 CFU/plate. Conventional methods for isolation and identification of microorganisms were used (Quinn et al. 1994).

Mycology
Samples from 233 of the 239 mares were also cultured for fungi on sabouraud dextrose agar 2% (Difco laboratories, Detroit, M) with chloramphenicol (0.5 µg/ml, Fluka Chemi, Buchs, Switzerland) and incubated at 30°C for 5 days. In the identification of fungus yeast was not identified to species except for Candida albicans according to McGinnis (1980). Antimicrobial susceptibility testing A microtiter plate system (VETMIC TM , SVA) was used for the antimicrobial susceptibility tests. The test was done according to the manufacturer's instruction. In brief, each well was inoculated with 50 µl of Mueller Hinton broth (Merck, KgaA, Darmstadt, Germany) to which 10 3 to 10 4 CFU of the bacteria to be tested were added. The wells were sealed with transparent adhesive tape and incubated at 35-37°C for 16-18 h. The lowest concentration of an antibiotic completely inhibiting bacterial growth was registered as Minimum Inhibitory Concentrations (MIC). Results were categorised by using the breakpoints for resistant, intermediate and sensitive recommended by the NCCLS for bacteria  , 1997).

Statistical analysis
The analysis of frequencies of the various fertility problems connected with microbiological diagnosis was made by chi-square analysis within the frequency procedure in SAS (SAS Inc., 1990).

Bacteriology
From 152 positive samples out of 239 sampled mares one or 2 significant bacterial species were isolated and identified. Ninety-two (38%) of the positive samples yielded growth of one single species in pure or almost pure culture. Thirty-one (13%) mares yielded 2 species and 29 (12%) mares yielded 1 species dominating on the agar plate together with sparse non-specific mixed culture. In samples from the mares without significant growth 57 (24%) yielded no growth at all and 30 (13%) growth of non-specific mixed culture. The bacterial species mostly isolated was E. coli, yielding 104 isolates, thereof 64 isolates in pure or almost pure culture (Table 1). When the bacterial growth was quantified, 72 E. coli isolates yielded abundant, 21 moderate and 11 sparse growth. From two mares, two different isolates of E. coli were isolated. Only 5 of the 104 E. coli isolates were haemolytic E. coli, 3 of these in pure culture. When the bacterial growth was quantified, 3 of these haemolytic E. coli yielded abundant, 1 moderate and 1 sparse growth. The second most frequently isolated species was ß-haemolytic streptococci yielding 31 isolates, thereof 12 grew in pure culture (Table 1). When the bacterial growth was quantified, 12 Streptococcus isolates yielded abundant, 12 moderate and 7 sparse growth. 1 When the MIC value was above the range tested, the value for the next titration step (the value just above the range) was used. 2 All isolates equal to or less than 1 were regarded as S since this was the lowest concentration tested. 3 All isolates equal to or less than 0.5 were regarded as S since this was the lowest concentration tested. 4 The vertical line shows the break point between S and Rt, no I-sensitivity is given. 5 The MIC value for trimethoprim tested in combination with sulfamethoxazol (1:20) is given. NT = not tested, the titration step is not included in the VetMIC TM system.

Relation between fertility problems and microbiological diagnosis
From repeat breeding mares, as well as from mares with clinical symptoms of endometritis, E. coli was the most frequently isolated species. ß-haemolytic streptococci were more frequently (p<0.01) associated with clinical endometritis than with repeat breeding (68 versus 23% of ß-haemolytic streptococcal isolates). The opposite was true for E. coli (p<0.01) (38 versus 53% of E. coli isolates).

Mycology
From 15 mares yeast was isolated (13 of these further typed as not being Candida albicans) and from 1 mare mould was isolated. These were all in mixed culture with bacteria.

Antimicrobial susceptibility
Among the 104 E. coli isolates, resistance was most common to cephalothin, streptomycin, trimethoprim/sulphamethoxazole (TMP) and ampicillin (Table 2). Several isolates were resistant to more than one antimicrobial. Enrofloxacin was the only one of the 10 tested antimicrobial agents for which no resistance was noted. The 5 haemolytic E. coli isolates were all classified as susceptible to TMP and gentamicin (Table 2). Among the 31 ß-haemolytic streptococcal isolates resistance was most common to gentamicin, neomycin, oxytetracycline, and to TMP (Table 3). All isolates were classified as susceptible to the ß-lactam antibiotics, penicillin G and ampicillin, and also to cephalothin, erythromycin, spiramycin and chloramphenicol.

Discussion
Mares included in this study were selected for reproductive disorders and only 32% of them yielded no significant growth of microorganisms. This figure should be compared with studies where mares have not been selected for reproductive problems. In these studies 70% (Redaelli & Codazza 1977), 68% (Shin et al. 1979) and 61% (Ricketts et al. 1993) of the mares yielded no significant growth. Other possible reasons for the difference in outcome between studies could be that different breeds have been studied and/or there was different breeding management between the studied populations. Also the sampling technique and microbiological culture routines may influence the results. Our data suggest that E. coli is the microorganism most frequently associated with fertility problems in the mare and that ß-haemolytic streptococci are the second most frequent. Further, E. coli seems to be more associated with repeat breeding without clinical symptoms than with clinical symptoms of endometritis. The opposite relation seems to apply to ß-haemolytic streptococci. This finding may be of interest for clinical considerations. The overall dominance of E. coli relative to ß-haemolytic streptococci in uterine swab samples was consistent in both years of sampling. The dominance of E. coli is in contrast to studies from other countries, where ß-haemolytic streptococci have been the bacteria most commonly isolated. We cannot give a causal explanation to this dominance of E. coli in the present study. Other studies have mostly been performed in normal populations of mares (Shin et al. 1979, Ricketts et al. 1993, which may be one explanation of this difference. But also a study in barren mares (Langoni et al. 1997) shows this dominance of ß-haemolytic streptococci. Further, it has earlier been suggested that nonhaemolytic E. coli is a non-pathogen in the equine uterus (Barrelet 1995). In our study, 99 of the 104 E. coli isolates were non-haemolytic E. coli. Given the clinical history of these 99 mares, and the fact that most of the non-haemolytic isolates yielded abundant growth in pure or almost pure culture, we suggest that also non-haemolytic E. coli in the equine uterus may cause fertility problems. In the present study, fungi were the third most frequent microbiological finding. Yeasts or mould were always isolated together with bacteria. Unfortunately, it was not possible to tell whether fungi or bacteria caused the primary infection. Fungal infections of the non-pregnant equine uterus were earlier said to be uncommon (Redaelli & Codazza 1977), but more recently fungal infections have been believed to be more frequent, possibly due to the widespread use of antibiotics and the increasingly intensive management and manipulation of reproduction in mares (Blue 1987, Le Blanc 1997. Staphylococcus aureus is reported to be a rather frequently isolated species from the equine uterus in a normal population of mares (Ricketts et al. 1993, Shin et al. 1979. In our study this species was not isolated at all, only coagulase-negative Staphylococcus spp. were isolated from 3 mares. Notable in the present study is also that well-known uterine pathogens such as P. aeruginosa and K. pneumoniae were only isolated from one mare each. The bacterial species isolated may be influenced by the stud farm management and the breeding regime used. In Sweden as well as in this study, the two dominating breeds are the Swedish Warmblood and the Standardbred Trotter. These breeds are mainly bred by AI, 71% and 88% for the Swedish Warmblood (Hästavel, 1998) and the Standardbred Trotters (STC, 1999), respectively. Also the sampling technique influences the culture results. In the present study, most of the isolated E. coli and ß-haemolytic streptococci yielded moderate or abundant growth of the isolated bacteria, indicating that these isolates represented an infection in the uterus rather than a vulvovestibular contaminant (Hinrichs et al. 1988, Waelchli et al. 1992. In this study, samples were transported overnight and when a time-span of this kind exists between collections and cultures the choice of transport medium is likely to influence the culture results (Shin et al. 1979, Ricketts et al. 1993. A general global rise in antibiotic resistance has been linked to an increased use of antibiotics (Fox 1997, Swartz 1997. In stud farm practice antibiotics have long been used both prophylactically before breeding, as a treatment of endometritis (Kenney et al. 1975, Shin et al. 1979 as well as in semen extenders (Burns et al. 1975, Kenney et al. 1975). In the present study, resistance to several commonly used antimicrobials was recorded. Notably as much as 15% of the E. coli isolates were resistant to TMP and 4% of the isolates to gentamicin. With respect to gentamicin, McCue et al. (1991) report a sensitivity in only 86% of E. coli isolates from equine endometrial swabs collected in the US. Our corresponding figure is 96%. This difference might be due to differences in how often the drug is used. The distribution of MIC-values for resistant E. coli isolates is rather consistent when comparing isolates from different species and organs (SVARM 2001). All ß-haemolytic streptococcal isolates were uniformly sensitive to 6 of 11 tested antibiotics, which is in accordance with Shin et al. (1979). As expected, all streptococcal isolates were sensitive to ß-lactam antibiotics.