Uterine bacterial infections are an important cause of reduced fertility in mares.1–3 Endometrial infections are reported in 25% to 60% of barren mares,2 contributing to a major economic loss to the equine industry.2–4 Paramount to the successful treatment of bacterial endometritis is rapid recognition of the clinical signs, accurate diagnosis, and implementation of appropriate treatment. Proper identification of the causative agent and determining an appropriate treatment are important diagnostic steps, requiring correct interpretation of both microbiological and cytologic data.
Although the uterus probably does not have normal flora, bacteria may be isolated from 30% of swab samples, even when these are collected with appropriate precautions.5 However, when cultured immediately without enhancement medium, this growth is typically light and does not include recognized potential pathogens. Potentially pathogenic organisms, notably Streptococcus equi subsp zooepidemicus and Escherichia coli, reside in the caudal tract5 and may be introduced into the uterus through a variety of ways, including during breeding (both natural and artificial), urogenital veterinary examinations, or breakdown of physical barriers to infection.2,6 It has been demonstrated that the most common bacteria isolated from mares with uterine infections include S equi subsp zooepidemicus, E coli, Staphylococcus aureus, Klebsiella pneumoniae, Pseudomonas aeruginosa, Bacteroides fragilis, and Bacteroides ureolyticus.7,8 A recent study4 evaluated results of bacteriologic culture and cytologic evaluation and pregnancy rates; however, no studies have addressed susceptibility patterns of bacteria isolated from uteri of mares or their changes in resistance patterns over time.
In 2009, the USDA reported that the breeding industry in the state of Florida was the second greatest producer of horses in the United States9 having a $5.1 billion impact on the Florida economy.10 The objective of the study reported here was to describe the most common bacterial isolates associated with the reproductive tract in breeding mares in Florida, to determine the extent and patterns of resistance to routinely selected antimicrobials, and to examine the relationship between potential pathogens and uterine cytologic abnormalities. We hypothesized that the population characteristics and resistance pattern would be similar to published data in other geographic locations.7,8
Materials and Methods
This retrospective study was based on uterine microbiological culture and susceptibility testing of samples from mares of breeding age collected at an equine referral center as well as those submitted from referring practitioners in a 100-mile radius in central Florida during 2003 to 2008. Breeds represented were predominantly Thoroughbreds (41.4%) and Quarter Horses (19%), combined with other breeds (40%) including Missouri Foxtrotter, Paso Fino, Morgan, Arabian, warmblood, and draft breeds.9 Data collected for this study were limited to cytologic evaluation and bacterial culture and susceptibility results. Specific signalment and history data were not recorded for the mares for which samples were collected. Uterine samples were collected from horses during routine prebreeding reproductive examinations as well as part of the reproductive diagnostic workup for infertility; however, the proportion of samples collected for prebreeding or diagnostic examination was not determinable.
Samples were collected with double-guarded swabs, small-volume lavages, or biopsies or were submitted as preplated culture samples. Each sample was used to inoculate a blood agar plate and Levine eosinmethylene blue agar platesa within 6 hours after collection, with the exception of preplated samples, which were plated by practitioners < 12 hours prior to submission. Plates were inoculated and incubated in accordance with guidelines of the Clinical Laboratory Standards Institute11 under aerobic, humid atmospheric conditions at 37°C for up to 48 hours. The extent of growth of each culture was characterized as no growth, 0; minimal growth, 1; mild growth, 2; moderate growth, 3; or marked growth, 4. For the purposes of this study, a score ≥ 2 was considered a positive result for potential pathogenic growth.
Susceptibility testing was determined with the following antimicrobials: amikacin, ampicillin, cefazolin, enrofloxacin, gentamicin, penicillin, polymyxin B, oxytetracycline, trimethoprim sulfa, ticarcillin, and ticarcillin with clavulanic acid. Susceptibility of each organism to each drug was based on guidelines recommended by the Clinical Laboratory Standards Institute for veterinary pathogens.12 Organisms that were categorized as intermediate were also considered to be resistant for the purposes of this study. Thus, each isolate was given a binary score (0 = susceptible and 1 = resistant) for each drug.
Samples for which cytology was requested by the referring veterinarian were stained with a Romanowsky stain variant.b The presence of inflammation was scored 0 through 4 by 1 of 3 medical technicians on the basis of the mean quantitative neutrophil count in 10 hpfs (magnification, 100X) where 0 = none, 1 = occasional neutrophil, 2 = 2 to 3 neutrophils/hpf, 3 = 5 to 10 neutrophils/hpf, and 4 = > 10 neutrophils/hpf. For the purposes of this study, a score ≥ 2 neutrophils/hpf was considered indicative of endometritis.13
Statistical analysis—For those organisms considered pathogens, the percentage of isolates resistant to each drug was reported overall and for each year as mean, SD, and 95% confidence interval. The proportions of resistant isolates for all years to each drug were compared via a general linear model in statistical software14,c followed by post hoc evaluation via a Tukey test. To detect a significant difference of resistance among years for each antimicrobial, a multiple comparisons test for proportions was performed.d For inflammation, the percentage of isolates associated with each inflammatory score was determined for each year and each organism. To assess the likelihood that either E coli or S equi subsp zooepidemicus was associated with inflammation, each isolate was given a binary score of 0 if the inflammatory score was < 2 and 1 if the inflammatory score was > 2. An OR was then determined for the association with inflammation for all years for each organism. The percentage of samples associated with an inflammatory score of 2 or higher was compared for each organism (E coli and S equi subsp zooepidemicus) among years via χ2 analysise,f or Fisher exact test when indicated. χ2 Analysis was also used to determine whether there was a significant difference in resistance between cultures with an inflammatory response and those without. For all statistical comparisons, values of P < 0.05 were considered significant.
Results
Sample submission methods—In general, samples were collected by means of a double-guarded uterine technique in 95% or more of reported cases (n = 1,451). Alternative techniques were not used until 2007. In 2007, the only alternative method was preplated culture, which represented 12 (31%) sample submissions. The following year (2008), preplated cultures accounted for 9 (14.3%) sample submissions, low-volume lavage for 5 (7.8%), and uterine biopsy tissue for 2 (1.5%). Over the entire study period, the breakdown of techniques by organism was as follows: S equi subsp zooepidemicus-positive cultures were submitted as uterine swab (n = 128 [95%]), preplated culture sample (4 [2.9%]), low-volume lavage (2 [1.4%]), and uterine biopsy (1 [0.7%]). Cultures with positive E coli growth were submitted as uterine swab (n = 115 [96%]), preplated culture samples (2 [1.7%]), low-volume lavage (1 [0.8%]), and biopsy (1 [0.8%]). Where P aeruginosa was cultured, uterine swabs contributed to 34 (95%) cases, preplated culture samples contributed to 2 cases (5.6%), and low-volume lavage contributed to 2 cases (5.6%). Finally, K pneumoniae cultures were submitted on swabs (n = 23 [96%]) and preplated culture (1 [4%]). On the basis of the overwhelming number of double-guarded uterine swab samples submitted in comparison with other sampling techniques, these are not likely to have impacted the data of the present study.
Isolate description—Samples were collected from a total of 7,655 mares. A total of 8,296 uterine samples were collected during the 6-year study period. The number of samples submitted each year and the proportion positive for growth considered pathogenic for each year was as follows: 2003, 304 (43%); 2004, 637 (35%); 2005, 352 (22%); 2006, 645 (31%); 2007, 405 (27%); and 2008, 529 (37%). A significant difference could not be detected in the percentage of total isolates represented by each organism among years (Table 1; Figure 1). The positive growth samples that yielded growth of the different potential pathogens during the 6-year period were E coli (n = 729 [29%]), β-hemolytic S equi subsp zooepidemicus (733 [28%]), gram-negative Enterobacteriaceae (excluding E coli; 266 [10%]), P aeruginosa (141 [6%]), and K pneumoniae (81 [3%]). Nonbacterial organisms identified on culture included yeast (n = 62 [2%]) and other nonyeast fungal organisms (18 [1%]). A total of 496 (19%) cultures yielding bacterial growth grew only organisms considered nonpathogenic.
Percentages of organisms cultured from uterine samples collected from breeding mares (n = 2,576) from January 2003 to December 2008 at an equine referral center as well as from referring practitioners in a 100-mile radius in central Florida.
Citation: Journal of the American Veterinary Medical Association 242, 7; 10.2460/javma.242.7.977
Population phenotype of organisms cultured from uterine samples collected between 2003 and 2008 at an equine referral center as well as from referring practitioners in a 100-mile radius in central Florida (number of mares affected).
| Variable | 2003 | 2004 | 2005 | 2006 | 2007 | 2008 | Totals |
|---|---|---|---|---|---|---|---|
| Total cultures | 714 | 978 | 1,604 | 2,107 | 1,476 | 1,417 | 8,296 |
| No growth | 410 | 637 | 1,252 | 1,462 | 1,071 | 888 | 5,720 |
| Pathogenic growth | 304 | 341 | 352 | 645 | 405 | 529 | 2,576 |
| Streptococcus equi subsp zooepidemicus | 72 | 78 | 111 | 202 | 124 | 146 | 733 |
| Escherichia coli | 104 | 110 | 80 | 190 | 111 | 134 | 729 |
| Enterobacteriaceae | 45 | 22 | 34 | 72 | 42 | 51 | 266 |
| Pseudomonas aeruginosa | 6 | 13 | 10 | 53 | 21 | 38 | 141 |
| Staphylococcus spp | 27 | 20 | 26 | 13 | 12 | 32 | 130 |
| α-Streptococcus spp | 10 | 26 | 16 | 6 | 25 | 22 | 105 |
| Klebsiella pneumoniae | 4 | 3 | 7 | 31 | 12 | 24 | 81 |
Of the 2,576 potential pathogen-positive cultures, 324 (12.6%) samples yielded growth of ≥ 2 bacteria. Escherichia coli was the organism most commonly associated with culture of multiple potential pathogens (n = 137 [42% of multiple-growth cultures]). These isolates represented 19% of the total cultures identified with E coli growth throughout the 6-year study period. Of the 137 cases in which E coli was isolated in conjunction with another organism, the coisolates were S equi subsp zooepidemicus (n = 46 [34%]), K pneumoniae (20 [15%]), Enterobacteriaceae (14 [10%]), P aeruginosa (13 [9%]), or Bacillus spp (12 [9%]).
The second most common isolate associated with multiple potential pathogenic growth was S equi subsp zooepidemicus (n = 97 [30% of multiple-growth cultures]). These isolates represented 13% of the total cultures identified with S equi subsp zooepidemicus growth throughout the 6-year study period. Of the 97 cases in which S equi subsp zooepidemicus was isolated in conjunction with another organism, the coisolates were E coli (n = 42 [43%]), Enterobacteriaceae (16 [16%]), P aeruginosa (8 [8%]), nonfermenting Enterobacteriaceae (8 [8%]), and Staphylococcus spp (6 [6%]).
Resistance phenotypes—Those clinically relevant organisms (from published data7) for which a sufficient number of isolates (n > 75) in the sample population were available to assess antimicrobial resistance included E coli, S equi subsp zooepidemicus, Enterobacteriaceae spp, P aeruginosa, and K pneumoniae.
The resistance patterns for E coli isolates among the years sampled were summarized (Table 2). Among the isolated bacteria, E coli expressed the greatest amount of resistance (P < 0.05) on the basis of proportion of isolates resistant to trimethoprim-sulfonamide, ampicillin, oxytetracycline, polymyxin B, and ticarcillin. Resistance of E coli was least toward amikacin and enrofloxacin. Significant differences in the proportion of resistant E coli isolates could not be detected among years for amikacin, cefazolin, ceftiofur, oxytetracycline, and ticarcillin-clavulanic acid. However, the proportion of resistant isolates differed significantly (P < 0.05) among years for ampicillin, enrofloxacin, gentamicin, polymyxin B, trimethoprim sulfa, and ticarcillin. The individual changes for each antimicrobial were summarized (Table 2).
Percentage of E coli isolates (n = 532) collected from breeding mares resistant to antimicrobial drugs each year from January 2003 through December 2008.
| Year | AN | AM | CZ | XNL | ENO | GM | PB | T | SXT | TIC | TIM |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2003 (n = 55) | 9 | 42x,y | 35 | 27 | 20y,z | 22 | 95y | NA | 49 | 33x,y | NA |
| 2004 (n = 55) | 6 | 31x | 16 | 16 | 9x,y | 7 | 55y | NA | 44x | 18x | NA |
| 2005 (n = 62) | 18 | 45x,y | 27 | 15 | 3x | 31 | NA | 42 | 53 | 31x,y | NA |
| 2006 (n = 163) | 10 | 53y,z | 34 | 17 | 4x | 33 | 23x | 50 | 63 | 48y,z | 16 |
| 2007 (n = 78) | 22 | 67y,z | 27 | 13 | 22y,z | 23 | 22x | 49 | 73y,z | 44x,y | 21 |
| 2008 (n = 119) | 16 | 61y,z | 29 | 30 | 12x,y | 29 | 17x | 37 | 69y,z | 42x,y | 21 |
| Mean ± SD | 14 ± 6a | 50 ± 13b,c | 28 ± 7a,b | 20 ± 7a | 12 ± 8a | 24 ± 9a | 42 ± 33b,c | 44 ± 6b,c | 59 ± 12b,c | 36 ± 11a,b,c | 19 ± 3a,b |
| 95% CI | 9–18 | 39–60 | 23–33 | 14–25 | 5–18 | 16–31 | 16–69 | 38–51 | 49–68 | 27–45 | 16–22 |
AN = Amikacin. AM = Ampicillin. CZ = Cefazolin. ENO = Enrofloxacin. GM = Gentamicin. NA = Not available for year because this antimicrobial was not on panel for susceptibility testing. PB = Polymyxin B. SXT = Trimethoprim sulfonamide. T = Oxytetracycline. TIC = Ticarcillin. TIM = Ticarcillin with clavulanic acid. XNL = Ceftiofur.
For all years, mean proportion of isolates resistant to each drug with shared superscript letters did not differ significantly (P < 0.05) from one drug to another.
Mean proportion of isolates resistant to each drug with shared superscript letters did not differ significantly (P < 0.05) for that year, compared with another year.
The resistance patterns for β-hemolytic S equi subsp zooepidemicus isolates among the years sampled were summarized (Table 3). For β-hemolytic S equi subsp zooepidemicus (n = 596), a greater proportion of isolates were found to be resistant in 2003 and 2006, compared with all other years (P < 0.05). Resistance was greatest (P < 0.05) to oxytetracycline, enrofloxacin, and penicillin and least to ampicillin, gentamicin, cefazolin, ceftiofur, ticarcillin, and ticarcillin-clavulanic acid. A significant change in resistance could not be detected among years for gentamicin, oxytetracycline, trimethoprim-sulfonamide, or ticarcillin-clavulanic acid. However, the proportion of resistant isolates differed significantly among years for ampicillin, cefazolin, ceftiofur, enrofloxacin, penicillin, and ticarcillin (P < 0.05). The individual changes for each antimicrobial were summarized (Table 3).
Percentage of S equi subsp zooepidemicus isolates (n = 596) collected from breeding mares resistant to antimicrobial drugs each year from January 2003 through December 2008.
| Year | AM | CZ | XNL | ENO | GM | P | T | TIC | TIM |
|---|---|---|---|---|---|---|---|---|---|
| 2003 (n = 59) | 34y,z | 25y,z | 15x,z | 88z | 17 | 52z | NA | 12x,z | NA |
| 2004 (n = 51) | 25x,z | 7x | 5x | 88z | 14 | 53z | NA | 6x,y | NA |
| 2005 (n = 102) | 17x,z | 5x | 10x,z | 48y | 19 | 28x,z | 58 | 7x,y | NA |
| 2006 (n = 150) | 9x | 6x | 4x | 3x | 15 | 11w | 43 | 2x | 5 |
| 2007 (n = 108) | 13x | 7x | 5x | 34y | 13 | 14w,x | 51 | 11y,z | 6 |
| 2008 (n = 126) | 15 | 9x | 17y,z | 21y | 13 | 19x | 52 | 15y,z | 7 |
| Mean ± SD | 19 ± 9a | 10 ± 8a | 9 ± 6a | 47 ± 35 | 15 ± 2a | 29 ± 19a,b | 51 ± 6b | 9 ± 5a | 6 ± 1a |
| 95% CI | 11–26 | 4–16 | 5–14 | 19–75 | 13–17 | 14–45 | 45–57 | 5–13 | 4–7 |
Means with different symbols were significantly different across years.
P = Penicillin.
For all years, proportion of isolates resistant to each drug with shared superscript letters did not differ significantly (P < 0.05) from one drug to another.
Proportion of isolates resistant to each drug with shared superscript letters did not differ significantly (P < 0.05) for that year, compared with another year.
See Table 2 for remainder of key.
The resistance patterns for Enterobacteriaceae (n = 236) isolates among the years sampled were summarized (Table 4). The proportion of resistant Enterobacteriaceae isolates was greatest for penicillin, ampicillin, and cefazolin and least for amikacin and enrofloxacin (P < 0.05). Significant changes in resistance could not be detected among years for amikacin, ceftiofur, or penicillin. However, resistance did differ among years for ampicillin, cefazolin, enrofloxacin, gentamicin, polymyxin B, oxytetracycline, trimethoprim-sulfonamide, ticarcillin, and ticarcillin-clavulanic acid (P < 0.05). The individual changes for each antimicrobial were summarized.
The resistance patterns for P aeruginosa (n = 133) isolates among the years sampled was summarized (Table 5). The proportion of resistant P aeruginosa isolates was greatest for enrofloxacin, ticarcillin, and gentamicin and least for amikacin (P < 0.05). Significant differences could not be detected among years for amikacin, gentamicin, or ticarcillin. The small sample population coupled with variability among years precluded statistical analysis of changes in resistance among years for P aeruginosa.
Percentage of Enterobacteriaceae isolates (n = 236) collected from breeding mares resistant to antimicrobial drugs each year from January 2003 through December 2008.
| Year | AN | AM | CZ | XNL | ENO | GM | P | PB | T | SXT | TIC | TIM |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2003 (n = 45) | 34 | 90$y | 79x,y | 59 | 17x | 38xy | 100 | 97y | NA | 34x | 55y | NA |
| 2004 (n = 22) | 14 | 43x | 71w | 43 | 14x | 14x | 86 | 100y | NA | 0 | 14x | NA |
| 2005 (n = 27) | 11 | 78y,z | 44x,y | 33 | 11xy | 41y | 96 | NA | 56 | 30x,y | 48y | NA |
| 2006 (n = 64) | 24 | 74$y | 84x,y | 30 | 11x | 49y | 89 | 20w | 45 | 72x,y | 52y | 19x |
| 2007 (n = 37) | 14 | 73y,z | 54x,y | 32 | 41y | 19x,y | 92 | 38y | 49 | 59y,z | 35y | 24x,y |
| 2008 (n = 41) | 24 | 88z | 80z | 41 | 17x | 27x,y | 93 | 32x | 44 | 61y,z | 56y | 41y |
| Mean ± SD | 20 ± 9a | 74 ±17d,e | 68 ±14d | 40 ±11c | 19 ±11a | 31 ±14a,b,c | 94 ± 5e | 57 ±38c,d | 48 ±5c,d | 43 ±26c,d | 43 ±16c,d | 28 ±12a,b,c |
| 95% CI | 13–27 | 61–88 | 56–79 | 31–48 | 10–27 | 20–42 | 89–98 | 24–91 | 43–53 | 21–64 | 30–56 | 15–42 |
For all years, mean proportion of isolates resistant to each drug with shared superscript letters did not differ significantly (P < 0.05) from one drug to another.
Proportion of isolates resistant to each drug with shared superscript letters did not differ significantly (P < 0.05) for that year, compared with another year.
See Tables 2 and 3 for remainder of key.
Percentage of P aeruginosa isolates (n = 133) collected from breeding mares resistant to antimicrobial drugs each year from January 2003 through December 2008.
| Year | AN | ENO | GM | TIC |
|---|---|---|---|---|
| 2003 (n = 6) | 17 | 100 | 33 | 17 |
| 2004 (n = 10) | 20 | 90 | 10 | 20 |
| 2005 (n = 10) | 10 | 40 | 0 | 20 |
| 2006 (n = 51) | 10 | 12 | 41 | 33 |
| 2007 (n = 20) | 30 | 55 | 40 | 40 |
| 2008 (n = 36) | 31 | 44 | 42 | 42 |
| Mean ± SD | 20 ± 9a | 57 ± 33b | 28 ± 18a | 29 ± 11a |
| 95% CI | 12–27 | 30–83 | 13–42 | 20–37 |
Because of the low number of samples, statistical analysis within drugs could not be performed.
Means with different superscript letters were significantly different across years.
See Table 2 for remainder of key.
Finally, the resistance pattern for K pneumoniae (n = 78) isolates among the years sampled was summarized (Table 6). Resistance was greatest in 2008 in comparison with all other years. The proportion of resistant K pneumoniae isolates was greatest for ticarcillin and cefazolin and least for enrofloxacin (P < 0.05). Because of the small sample population and high level of variability, it was not possible to perform statistical analysis evaluating a change in resistance among years.
Percentage of K pneumoniae isolates (n = 133) collected from breeding mares resistant to antimicrobial drugs each year from January 2003 through December 2008.
| Year | AN | CZ | XNL | ENO | GM | T | SXT | TIC | TIM |
|---|---|---|---|---|---|---|---|---|---|
| 2003 (n = 4) | 25 | 75 | 50 | 0 | 0 | NA | 50 | 75 | NA |
| 2004 (n = 2) | 0 | 50 | 0 | 0 | 0 | NA | 0 | 50 | N |
| 2005 (n = 6) | 0 | 17 | 0 | 0 | 17 | 0 | 17 | 67 | NA |
| 2006 (n = 30) | 3 | 57 | 33 | 3 | 30 | 47 | 53 | 63 | 30 |
| 2007 (n = 12) | 8 | 42 | 17 | 25 | 17 | 8 | 25 | 58 | 8 |
| 2008 (n = 24) | 26 | 57 | 39 | 30 | 43 | 43 | 57 | 74 | 48 |
| Mean ± SD | 10 ± 12a | 49 ± 19 | 23 ± 21a | 10 ± 14a | 18 ± 17a | 25 ± 24a,b | 34 ± 23a,b | 65 ± 10 | 29 ± 20a,b |
| 95% CI | 1–20 | 34–65 | 6–40 | –1 to 21 | 4–31 | 1–48 | 15–52 | 57–72 | 6–51 |
See Tables 2 and 5 for remainder of key.
Cytologic examination—Of the 2,872 samples evaluated for inflammation, 2,270 (79%) had a score ≥ 1. Inflammation was considered substantial (score > 2 cells/hpf) in 430 (19%) evaluated samples.13 Bacteria were isolated in 50% of samples classified as negative for inflammation and in 50% of samples classified as positive for inflammation.
In samples negative for inflammation (score < 2 cells/hpf), the proportions of isolated bacteria were as follows: E coli, 113 of 197 (57%); S equi subsp zooepidemicus, 73 of 193 (38%); P aeruginosa, 29 of 53 (55%); K pneumoniae, 14 of 25 (56%); and Enterobacteriaceae, 39 of 74 (53%). In samples positive for inflammation (score > 2 cells/hpf), the proportions of isolated bacteria were as follows: E coli, 84 of 197 (43%); S equi subsp zooepidemicus, 120 of 193 (62%); P aeruginosa, 24 of 53 (45%); K pneumoniae, 11 of 25 (44%); and Enterobacteriaceae, 35 of 74 (47%). The incidence of having a combined positive bacterial culture with evidence of inflammation (> 2 inflammatory score) for the top 5 pathogens in this population of mares (E coli, S equi subsp zooepidemicus, P aeruginosa, K pneumoniae, and Enterobacteriaceae) was 274 of 544 (50%). Specifically, in samples positive for E coli, 84 (43%) were associated with an inflammatory score ≥ 2; however, cultures positive for S equi subsp zooepidemicus (n = 120) were associated with inflammation in 62% of samples, which was significantly (P < 0.05) greater than the percentage for E coli (Figure 2). There was no significant difference in resistance between the samples that had confirmed evidence of inflammation (via cytologic evaluation) and those without.
Frequency distribution of endometrial isolates associated with inflammation scores among years for Escherichia coli (A) or Streptococcus equi subsp zooepidemicus (B) collected from breeding mares. The presence of inflammation was scored 0 through 4 on the basis of mean quantitative neutrophil (PMN) count, where 2 = 2 to 3 neutrophils/hpf, 3 = 5 to 10 neutrophils/hpf, and 4 = > 10 neutrophils/hpf. A score ≥ 2 was considered indicative of endometritis.
Citation: Journal of the American Veterinary Medical Association 242, 7; 10.2460/javma.242.7.977
Discussion
The results of the present study, which evaluated 8,296 samples in 7,665 mares over a 5-year period (2003 through 2008), indicated that, on the basis of susceptibility data, E coli and S equi subsp zooepidemicus are most commonly associated with endometritis, and antimicrobials currently recommended for empirical treatment of endometritis are appropriate. However, the results of this study also suggest that of the 2 isolates, S equi subsp zooepidemicus may be more likely to be a pathogen on the basis of the inflammatory response associated with its isolation.
Historically, multiple studies4,8,13–16 have evaluated culture techniques, cytologic findings indicative of endometrial inflammation, and the association of these assumed abnormalities on pregnancy rates in mares. Previous studies performed in other geographic areas have reported E coli, β hemolytic S equi subsp zooepidemicus, P aeruginosa, and K pneumoniae as being associated with cytologically or histologically confirmed endometritis.7,8 Historically, E coli and β hemolytic S equi subsp zooepidemicus have been associated with 50% to 80% of all bacterial endometritis cases in published data.17 The results of the present study support the population description that has been published.
Among the critical aspects of diagnosing equine endometritis is distinguishing infection from sample contamination by the environment, external genitalia, or vagina.1–4 A key criterion for the diagnosis of infection is the presence of inflammation combined with positive results of microbial culture. In the present study, the association of S equi subsp zooepidemicus with cytologic evidence of inflammation (62% of cases) supports this organism as a potential pathogen. The role of E coli isolates is less clear because the proportion of isolates associated with inflammation was only 43%. The presence of inflammation for both organisms may support either as a potential pathogen. However, E coli might also be considered a contaminant, meaning it is present as a nonpathogen.18 Its role under such circumstances needs further investigation, and it might not necessarily be innocuous. Our findings in regard to the relationship between inflammation and pathogenicity are in concert with other investigators. For example, other investigators have found the incidence of concurrent inflammation to be 65% for S equi subsp zooepidemicus and 55% for E coli.4,8
In our study, 29% of isolates were E coli, a figure that differs from previous studies7,18 finding 18% and 67% of isolates positive for E coli. However, the relevance of these older studies to the population of the present study is difficult to assess because they were in geographically distant locations (Italy and Sweden). This variability may be attributed to population of mares sampled, geographic location, and exposure of mares to various antimicrobials. In 2 studies4,19 performed in Lexington, Ky, with a similar population of breeding mares, isolates were positive for E coli in 33.5% and 42.2% of cases.
The most important contribution of our study is the focus on antimicrobial resistance. Thus far, minimal evidence has been provided for antimicrobial resistance phenotypes of bacteria within a reproductively active equine population in the United States. In our study population of potential bacterial pathogenic phenotypes, significant changes in resistance patterns were present among the different years for each organism. For E coli isolates, significantly decreased resistance was appreciated for polymyxin B, which is most likely due to a decreased number of practitioners who select this antimicrobial for use in practice. On the other hand, the proportion of isolates resistant to ampicillin, enrofloxacin, gentamicin, trimethoprim sulfa, and ticarcillin increased. The increased resistance may reflect increased use of these drugs by practitioners for intrauterine infusion (ticarcillin)20 or systemic treatment for placentitis (gentacin and trimethoprim-sulfonamide).21
Interestingly, for S equi subsp zooepidemicus, resistance to ampicillin, cefazolin, ceftiofur, enrofloxacin, and penicillin was significantly decreased. This decrease in resistance across years may be due to practitioner selection of antimicrobials on the basis of severity of clinical signs, susceptibility results, and antimicrobial treatment by clinicians. There was a significant increase in resistance to ticarcillin in 2007 and 2008, which is most likely due to a rise in popularity of this antimicrobial in recent years.
On the basis of these patterns of resistance versus susceptibility, this study suggests that the preferred empirical choices for treatment of pathogenic E coli are amikacin and enrofloxacin. These are both concentration-dependent drugs. Aminoglycosides are generally administered locally (intrauterine),22 thus assuring the high concentrations desirable for a concentration-dependent drug. However, enrofloxacin cannot be administered locally because of the extremely basic pH (8.9 to 10.9)23 of the commercial preparation. Yet other investigators have demonstrated that systemic administration of enrofloxacin (5 mg/kg, IV) results in drug concentrations (including the metabolite ciprofloxacin) in the endometrium that should be bacteriocidal toward susceptible bacteria.24 Compared with 2 studies7,18 of resistance patterns of E coli in sample populations, the present study showed a similar resistance pattern to the study of Albihn et al18 of bacterial culture in mares in Sweden. Interestingly, there was dramatically less detectable resistance in our study, compared with the study of Frontoso et al7 of bacterial culture in mares in Italy and antimicrobial susceptibility testing for amikacin (11%, compared with 48%) and oxytetracycline (12%, compared with 44%). This may be due to breeding population, environmental issues, or antimicrobial selection of practitioners.
Cultures that were positive for growth of S equi subsp zooepidemicus in the present study represented 28% of positive culture growth, which also differs from the retrospective studies of Frontoso et al7 (32%) and Albihn et al18 (14%). In the studies4,19 performed in Lexington, Ky, the percentage of cultures that were positive with growth ranged from 21% to 80%.
For β-hemolytic S equi subsp zooepidemicus, resistance was least toward ticarcillin, with or without clavulanic acid, ceftiofur, cefazolin, gentamicin, and ampicillin in this study. With the exception of gentamicin, each is a β-lactam antimicrobial. These have been used as intrauterine treatment because of their potent bactericidal nature and wide margin of safety.17,25,26 In this population of S equi subsp zooepidemicus–positive cultures, the percentage of isolates resistant to ticarcillin versus ticarcillin with clavulanic acid differed, albeit not significantly. The lack of significant difference probably reflects the small sample sizes but should not be interpreted to mean that the resistant rates are similar. In treatment of other conditions, the β-lactamase activity provided by clavulanic acid enhances efficacy by reducing resistance.20,26 Resistance may not be appreciated owing to minimal exposure of this population to the drug, insufficient subject population, or time sampled. However, because of market availability at this time in the United States, practitioners are unable to purchase ticarcillin alone.
The results of our study were again consistent with the findings of Albihn et al,18 who reported resistance patterns of S equi subsp zooepidemicus, with the following exceptions: resistance was less to gentamicin (15%, compared with 52%) but was greater to penicillin (29%, compared with 0%) and oxytetracycline (51%, compared with 23%). Interestingly, our resistance pattern was similar to the study of Frontoso et al7 of bacterial culture and antimicrobial susceptibility testing in mares in Italy, with the exception that we found less resistance for gentamicin (15%, compared with 65%).
Among the limitations of the present study was the lack of historical data regarding the reason for mare examination (ie, prebreeding exam vs reproductive dysfunction). The availability of this information would have facilitated identifying a relationship between organism growth and clinical signs rather than limiting our assessment to the relationship between an organism's growth and inflammation. We would have been able to further examine relationships between disease and resistance.
Our findings regarding isolated organisms were consistent with those of previous investigators in population characteristics and in that S equi subsp zooepidemicus is more likely than E coli to be associated with cytologic findings consistent with pathological changes. Our study was not able to detect an increase in proportion of equine uterine pathogens as a whole; however, there was a significant change in resistance among years for chosen antimicrobials. These findings support the importance of susceptibility testing as guidelines for treatment.
Becton Dickinson, Franklin Lakes, NJ.
Diff Quick, Hemal Stain Co Inc, Danbury, Conn.
PROC GLM, SAS, version 9.1, SAS Institute Inc, Cary, NC.
COMPROP macro, SAS, version 9.1, SAS Institute Inc, Cary, NC.
PROC FREQ, SAS, version 9.1, SAS Institute Inc, Cary, NC.
PROC LOGISTIC, SAS, version 9.1, SAS Institute Inc, Cary, NC.
References
1. Asbury AC. Endometritis in the mare. In: Morrow DA, eds. Current therapy in theriogenology diagnosis, treatment and prevention of reproductive diseases in small and large animals. 2nd ed. Philadelphia: WB Saunders Co, 1986;718–722.
2. Causey RC. Making sense of equine uterine infections: the many faces of physical clearance. Vet J 2006; 172: 405–421.
3. Asbury AC. Infectious causes of infertility. In: McKinnon AO, Voss JL, ed. Equine reproduction. Ames, Iowa: Blackwell Publishing, 1993;381–395.
4. Riddle WT, LeBlanc MM, Stromberg AJ. Relationships between uterine culture, cytology and pregnancy rates in a Thoroughbred practice. Theriogenology 2007; 68: 395–402.
5. Hinrichs K, Cummins MR, Sertich PL, et al. Clinical significance of aerobic bacterial flora of the uterus, vagina, vestibule, and clitoral fossa of clinically normal mares. J Am Vet Med Assoc 1988; 193: 72–75.
6. Causey RC. Uterine therapy for mares with bacterial infections. In: Samper JC, Pycock JF, McKinnon AO, ed. Current therapy in equine reproduction. St Louis: Elsevier, 2007;105–115.
7. Frontoso R, De Carlo E, Pasolini MP, et al. Retrospective study of bacterial isolates and their antimicrobial susceptibilities in equine uteri during fertility problems. Res Vet Sci 2008; 84: 1–6.
8. LeBlanc MM, Causey RC. Clinical and subclinical endometritis in the mare: both threats to fertility. Reprod Domest Anim 2009; 44(suppl 3): 10–22.
9. Florida Agriculture Center and Horse Park Authority. Economic and fiscal impact of the Florida horse park upon Marion county and the state of Florida. January, 2008. Available at: www.ocalacc.com/ocala_florida/articlefiles/338-Florida%20Horse%20Park%20Economic%20Impact%20Study.pdf. Accessed Feb 1, 2010.
10. Florida Department of Agriculture and Consumer Services. Equine horse industry. Available at: www.florida-agriculture.com/news/02-17-09.htm. Accessed Apr 9, 2009.
11. Clinical Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing. 16th informational supplement M100–S16. Wayne, Pa: Clinical Laboratory Standards Institute, 2005.
12. Clinical Laboratory Standards Institute. Methods for determining bactericidal activity of antimicrobial agents. Approved guideline M26-A. Vol 19. Wayne, Pa: Clinical Laboratory Standards Institute, 1999.
13. Dascanio JJ. Endometrial cytology. In: Frazer, GS, ed. Current therapy in equine medicine. 5th ed. Philadelphia: WB Saunders Co, 2003;226–228.
14. SAS/IML 9.1, user's guide: statistics. Cary, NC: SAS Institute Inc, 2004;63–234.
15. Card C. Post-breeding inflammation and endometrial cytology in mares. Theriogenology 2005; 64: 580–588.
16. Nielsen J, Troedsson M, Pedersen M, et al. Diagnosis of endometritis in the mare based on bacteriological and cytological examinations of the endometrium: comparison of results obtained by swabs and biopsies. J Equine Vet Sci 2010; 30: 27–30.
17. LeBlanc MM. The current status of antibiotic use in equine reproduction. Equine Vet Educ 2009; 21: 156–167.
18. Albihn A, Baverud V, Magnusson U. Uterine microbiology and anti-microbial susceptibility in isolated bacteria from mares with fertility problems. Acta Vet Scand 2003; 44: 121–129.
19. LeBlanc MM, Magsig J, Stromberg A. Use of a low volume uterine flush diagnosing endometritis in chronically infertile mares. Theriogenology 2007; 68: 403–412.
20. Van Camp SD, Papich MG, Whitacre MD. Administration of ticarcillin in combination with clavulanic acid intravenously and intrauterinely to clinically normal oestrous mares. J Vet Pharmacol Ther 2000; 23: 373–378.
21. Macpherson ML. Diagnosis and treatment of equine placentitis. Vet Clin North Am Equine Pract 2006; 22: 763–776.
22. Lu KG, Morresey PR. Reproductive tract infections in horses. Vet Clin North Am Equine Pract 2006; 22: 519–552.
23. Enrofloxacin material safety data sheet. Leverkusen, Germany: Bayer Heathcare LLC, 2004.
24. Papich MG, Van Camp SD, Cole JA, et al. Pharmacokinetics and endometrial tissue concentrations of enrofloxacin and the metabolite ciprofloxacin after i.v. administration of enrofloxacin to mares. J Vet Pharmacol Ther 2002; 25: 343–350.
25. Siu LK. Antibiotics: action and resistance in gram-negative bacteria. J Microbiol Immunol Infect 2002; 35: 1–11.
26. Dowling PM. Antimicrobial therapy. In: Bertone JJ, Horspool LJ, eds. Equine clinical pharmacology. Edinburgh: Saunders, 2004;13–47.
