Relationships between Pregnancy Rates, Uterine Cytology, and Culture Results in a Thoroughbred Practice in Central Kentucky

Mares that had inflammation on cytological specimens had lower 28-day pregnancy rates than mares with normal cytological findings, irrespective of culture results (p < 0.001). Day 28 pregnancy rates were also decreased in mares that had bacteria isolated from their uterus, even if cytological findings were normal (p < 0.001). Both uterine cytology and uterine culture were diagnostic; however, almost twice as many mares (1.8 times) were identified by cytological assessment than by uterine culture.

1. Introduction

In the central Kentucky Thoroughbred industry, all barren and maiden mares on their first breeding are required to have a clean uterine culture before mating; on the majority of farms, all mares (barren, maiden, and foaling) are required to have a clean culture on subsequent matings. Only mares that have a signed clean culture certificate are allowed to be mated. However, it is not always possible to wait 48 - 72 h to ensure that no bacteria are isolated from the swab before the mare is bred. Evaluation of cytological specimens in conjunction with uterine swabs has been promoted since the early 1980s as a technique that will enhance the diagnostic accuracy of culture results and can be used to identify mares early in estrus that have uterine inflammation [1-5]. Early studies by Knudsen [6] showed good correlation between the presence of neutrophils in cytological specimens and the isolation of bacteria when a guarded uterine swab was used to collect the sample. Many reports have confirmed these initial findings; however, no correlations between cytological findings, culture results, and pregnancy rates have been reported.

In this study, we determined the relationships between cytological findings, culture results, and 28-day pregnancy rates in a large group of Thoroughbred mares in central Kentucky. We hypothesized that pregnancy rates would be decreased in mares that had moderate or severe inflammation on cytological specimens. Furthermore, isolation of bacteria from uterine swabs would also result in decreased pregnancy rates if the cytological specimen had moderate or severe inflammation.

2. Materials and Methods

Uterine cytology and culture swabs were collected from 970 Thoroughbred mares during the 2001 - 2004 breeding seasons. Two populations of mares were used. The first group consisted of samples that were routinely collected from Thoroughbred mares that resided on farms that the first two authors visited daily. Uterine cytological specimens were collected concurrently with culture swabs from all mares on these selected farms with the exception of maiden mares who did not necessarily have a cytology performed before their first mating. The second population of mares consisted of mares that were repeat breeders and/or suspected to have inflammation or bacterial endometritis. These samples were taken by three ambulatory clinicians within the practice. Twenty-eight day pregnancy rates were obtained from the veterinarian's reproductive records or the billing records of the hospital.

Uterine cytology and culture specimens were collected with a Kalajian swab [a]. The culture instrument was either passed through a sterile vaginal speculum into the cervix and uterus or manually inserted by a veterinarian wearing a sterile sleeve. Once the instrument was in the uterus, the inner rod was pushed forward into the uterus to pop open the cap and the cotton swab on the end of the rod rotated for 10 - 15 s. The cotton swab was drawn back into the outer sheath, and the outer sheath containing the cap was twirled for 10 - 15 s to obtain cells within the cap. The swab was immediately placed in a charcoal medium [b]. Two methods were used to prepare the slides for microscopic examination. (1) The cap was tapped on a glass microscope slide. If the secretions placed on the glass slide were thick, a second slide was placed on the end of the first slide at a 45° angle and slid the length of the slide to distribute the secretions in a manner similar to making a blood smear. (2) A smear was made from the cap on a slide. The slide containing the endometrial cells was either fixed with a fixative spray [c] or air dried.

Uterine cultures were plated on blood and Levin eosin-methylene blue plates at the medical laboratory at Rood and Riddle Equine Hospital within 6 h of sample procurement. The plates were examined every 24 h for 3 days for the presence of bacteria or yeast. Bacteria were identified with BBL crystals. Antimicrobial sensitivities were determined for all organisms with the exception of Bacillus and Micrococcus, which were considered to be non-pathogens. Primary identification of yeast were determined by dark field. Further identification of specific yeast or fungal cultures was performed by the Microbiology Laboratory at Cornell University.

Cytological specimens were stained with Diff Quick stain [d]. Cytological specimens were evaluated by one of three medical technologists at x40 and under oil (x100). A minimum of 10 fields was evaluated for each slide. Findings were categorized as follows:normal cytology, epithelial cells and zero to two neutrophils per x100; moderate inflammation, epithelial cells and two to five neutrophils per x100; severe inflammation, epithelial cells and more than five neutrophils per x100; hypocellular specimen, no neutrophils or epithelial cells were present on the slide.

Breeding management of mares was determined by the attending veterinarian. Mares were usually bred only once before ovulation. They received an ovulatory agent, either human chorionic gonadotropin (hCG) [e] or Ovuplant [f]. Treatment after breeding varied among veterinarians. If cytological specimens were normal, some veterinarians infused antibiotics into the uterus at the time of the ovulation check (usually the day after mating), whereas others only gave oxytocin or no treatment. If cytological specimens exhibited inflammation, or pathogenic bacteria were isolated, mares were treated aggressively after breeding in most cases. Treatment included uterine lavages, intra-uterine antibiotics, and ecbolic drugs. In some cases, if bacteria were isolated and cytology results were normal, mares were only infused with either the appropriate antibiotic or a broad-spectrum antibiotic at the time of the ovulation check.

All p values were computed using Pearson's χ2 statistic. A p value of < 0.05 was considered significant.

3. Results

Fifteen hundred thirty-seven of the 2123 (72.4%) paired cytology and culture samples evaluated were normal (zero to two neutrophils and no bacteria isolated). Seventy-nine cytological specimens were hypocellular. These samples were not included in the final data analysis. Four hundred twenty-three cytological specimens had moderate or severe inflammation, whereas 231 uterine swabs had bacteria isolated. Sixty percent of the mares that had normal cytological specimens and no bacteria isolated from uterine swabs were pregnant at 28 days (controls). Mares with moderate or severe inflammation on cytological specimens had significantly lower 28-day pregnancy rates, irrespective of culture results (p < 0.01), compared with control mares. Mares with severe inflammation on cytological specimens had lower pregnancy rates than mares with moderate inflammation (p < 0.01). Bacteria were isolated from 19.7% of the cultures obtained from mares with moderate inflammation on cytological specimens and from 49% of mares with severe inflammation. Isolation of bacteria from the uterus of mares that had moderate or severe inflammation on cytological specimens did not further decrease pregnancy rates. However, isolation of bacteria in mares with no inflammation on cytological specimens was associated with decreased pregnancy rates compared with control mares (p < 0.01).

Eighty-one percent of the uterine swabs from which bacteria were isolated grew the following (listed in order of highest number of isolates): β-hemolytic streptococcus, E. coli, Pseudomonas, a non-pathogen, or two or more organisms. Inflammation was present in 65% of the cytological specimens obtained from mares that had endometritis because of β-hemolytic streptococcus and in 50 - 54.5% (range) of those mares from which E. coli, Pseudomonas, or a non-pathogen was isolated. If two or more organisms were isolated from the uterus, all paired cytological specimens had moderate or severe inflammation. The 28-day pregnancy rates of mares that had a β-hemolytic streptococcus, E. coli, Pseudomonas, or a non-pathogen isolated from their uterus ranged from 22% to 44%, with the lowest pregnancy rates seen in mares from which non-pathogens were isolated.

4. Discussion

Both uterine cytology and culture results were diagnostic in this retrospective study, because pregnancy rates were decreased in mares that had either uterine inflammation, bacteria recovered from their uterus, or both. Our data deviated from the dogma that bacteria isolated from the uterus is a contaminant or a non-pathogen if no inflammatory response is observed on uterine cytology. Pregnancy rates were significantly reduced in mares that had bacteria isolated from their uterus without concurrent inflammation on cytological specimen. Furthermore, the correlation between isolation of a single organism and inflammation on cytological specimens was not strong, because inflammation was present in only 50 - 65% of the samples from which a single organism was recovered. It may be that different bacteria induce different degrees of inflammation or that our method of procurement may not be diagnostic. However, only 79 of 2123 samples (3.7%) were hypocellular and did not contain adequate quantities of epithelial cells, indicating that viable samples were obtained with the sampling technique. It is possible that inflammatory cells may be restricted to an area within the uterus where bacteria are most abundant and that area may not be sampled with a culture rod. We have also observed a lack of correlation between cytological findings, uterine culture, and biopsy results when a small volume lavage is performed for uterine cytological specimen and culture [g]. Fluids collected from a small volume are more representative of uterine contents as the fluid contacts a larger area of the endometrium. We have found cytological specimens negative for inflammation even though a single pathogenic organism was isolated from the uterus of a repeat breeder who exhibited diffuse or focally moderate inflammation on uterine biopsy specimen. This dilemma needs further study.

Almost twice as many mares (1.8 times) were identified by cytological assessment than by uterine culture. By performing a uterine cytology, mares with uterine inflammation can be identified more quickly than by uterine culture because samples can be processed and interpreted by laboratory personnel or by the veterinarian the same day the sample is taken. Mares identified as having uterine inflammation can have additional diagnostics performed before breeding in an attempt to determine the source of the inflammation. These may include repeated ultrasonographic examinations for uterine fluid accumulation, an additional vaginal examination late in estrus to determine if there is vestibule-vaginal reflux, or a cervical examination to identify a cervical tear.

Mares that had bacteria isolated from uterine cultures were mated in some cases because culture results were negative at the 24- or 48-h examination, and bacteria were found on the 48- or 72-h examinations. Mares that had inflammation on cytological specimens without bacteria isolated many times would also be bred. Both groups of mares were treated aggressively after breeding with uterine lavage, ecbolics, intra-uterine antibiotics, and in some cases, systemic antibiotics. Pregnancy rates were similar within groups but were significantly lower than mares with normal cytologies and no bacteria. It is not known if these mares would have conceived without treatment.

Mares that are bred with uterine inflammation or bacteria isolated from their uterus may conceive and abort as late as 120 days of gestation [7]. We may have experienced additional losses between 28 and 120 days; however, we used 28-day pregnancy rates to avoid including any pregnancy loss that may have been caused by mare reproductive loss syndrome (MRLS) in 2001 and 2002. Mares typically lost their pregnancies after 38 - 40 days from MRLS.

Pregnancy rate in mares from which either Bacillus or Micrococcus (considered to be non-pathogens) were isolated from uterine swabs was only 22%. These mares would not have been treated aggressively after breeding. Because the pregnancy rates were decreased in these mares, a re-evaluation of our determination of pathogenic versus non-pathogenic bacteria is necessary.

In conclusion, we found that both uterine cytology and uterine culture were diagnostic in identifying endometritis. Inflammation on cytological specimen or bacteria isolated from the uterus was associated with a decrease in pregnancy rates; however, almost twice as many mares were identified from uterine cytological specimens. Our findings contrast with the dogma that bacteria isolated from the uterus of mares with normal cytological findings is a non-pathogen because mares with these results had lower pregnancy rates than control mares.

We thank Melissa Bender and Joshua Maysig for assistance with data management.

Footnotes

1. Kalayjian Industries, Inc., Signal Hill, CA 90755.
2. Difco Transport Medium Amies; Becton, Dickinson and Co, Sparks, MD 21152.
3. Cytology Fixative Spray; US Biotech Corp., Webbville, KY 41180.
4. Diff Quik; Hemal Stain Co. Inc., Danbury, CT 06810.
5. Chorulon; Intervet Inc. , Millsboro, DE 19966.
6. Ovuplant; Fort Dodge, Wyeth, Madison, NJ 07940.
7. LeBlanc MM. Unpublished observations. 2005.