The Veterinary Record, December 12, 1998
Uterine bacterial flora and uterine lesions in bitches with
cystic endometrial hyperplasia (pyometra)
G. K. Dhaliwal, C. Wray, D. E. Noakes
Veterinary Record (1998) 143, 659-661
Thirty-eight complete genital tracts were obtained fresh from
local veterinary practices after bitches had been ovario-
hysterectomised for the treatment of cystic endometrial
hyperplasia (pyometra). The aerobic and anaerobic culture of
swabs obtained aseptically from the uterine lumen demon-
strated the presence of Escherichia coli in 28 of the 29 cases
from which bacteria were isolated and, from 25, a pure cul-
ture was obtained. Sixteen different serotypes were identified
from 24 genital tracts; 032K+ CNF+ was isolated from four of
16 cases and 04K-; from three of 16 cases. Cytotoxin necrotis-
ing factors (CNFS) were identified in seven of the 16 serotypes;
they included 02, 04, 06, 022, 025, 045 and 075, which except
for 06 and 045 were haemolytic. Strains 04K- and 07K+ com-
pletely destroyed the integrity of the endometrial epithelium
whereas 088K+ and 088K- had the least effect. The scores for
the reduction in the integrity of the endometrial epithelium
and the degree of infiltration of inflammatory cells were
greater in the presence of CNF 1.
CYSTIC endometrial hyperplasia or pyometra is a disease of the
uterus which occurs most frequently in the luteal phase of the
oestrous cycle (metoestrus). In the absence of the use of exoge-
nous steroid hormones, which have been shown to predispose
bitches to the disease (Teunissen 1952, Dow 1959a), it can proba-
bly best be described as being associated with the abnormal
response of the bitch's uterus to ovarian hormones either with or
without a secondary bacterial infection (Hardy and Osborne
1974). In some bitches with pyometra the uterus is sterile
(Teunissen 1952) which suggests either that bacteria are not
always involved in the pathogenesis of the disease, or that they
were present initially but have been eliminated by the time
attempts to culture bacteria are made. When bacteria have been
isolated, Escherichia coli has been predominant (Dow 1958,
1959b, Grindlay and others 1973). Among the strains of E coli
which have been isolated from cases of pyometr4 in bitches, the
presence of the K-antigen was a common characteristic
(Sandholm and others 1975); in their study both the bladder and
uterus were infected by the same strains. A similar association
between the presence of the K-antigen and the virulence of uri-
nary tract infections has been observed in both the mouse and man
(Kalmanson and others 1975, McCabe and others 1975). Eight
strains of E coli have been isolated from cases of pyometra in
bitches, the most common being 04, 06, 02 and 075 in descending
order of frequency (Nomura 1984).
In this study, the bacterial flora of the uterus was studied in bitch-
es that had undergone ovariohysterectomy for the treatment of
pyometra, and an attempt was made to correlate the strains of
E coli isolated with the histopathological changes in the uterine wall.
Materials and methods
Source of material
Complete genital tracts were obtained from local veterinary
practices immediately after 34 bitches had undergone ovario-
G. K. Dhaliwal, DVM, MS, PhD, D. E. Noakes, BVetMed, PhD, FRCVS,
Department of Farm Animal and Equine Medicine and Surgery, Royal
Veterinary College, Hawkshead Lane, North Mymms, Hatfield AL9 7TA
C. Wray, BVM&S, PhD, MRCVS, FRCPath, Veterinary Laboratories
Agency, New Haw, Woodham Lane, Addlestone, Surrey KT15 3NB
659
hysterectomy for the treatment of pyometra. Some details of the
history and clinical signs of each case were recorded.
Microbiological examination
Using aseptic techniques, two swabs were taken from each
uterine horn, using a transport swab (Microbiological Supply
Company), before any tissues were sampled. They were cul-
tured aerobically and anaerobically. A duplicate sample was
immediately plated on MacConkey's agar (Lab Pak) for the
isolation of E coli. If growth occurred, five morphologically
different single colonies from the original plate were re-plated
and biochemical tests (API 20E test; Biomerieux) were done on
each colony to identify E coli positively. Positively identified
E coli cultures were then stored in a cryoprotectant medium
(Protect M vials; TSC) at -200C. The collection of five morpho-
logically different types of colonies from each uterus ensured
that most, if not all the types of E coli present in each case
were sampled. The colonies which were positively identified as
E coli were serotyped using the tube agglutination method
described by Orskov and others (1977). The presence of the
toxin cytotoxic necrotising factor (CNF) was recognised by its
cytopathic effects on verocells, as described by Caprioli
(1983).
Histopathological examination
Immediately after the swabs had been taken, whole thickness
samples (1 cm x 1 cm) of the uterine wall were obtained from the
middle of each horn. They were placed immediately in Mirsky's
fixative (National Diagnostics) and after sectioning, stained with
haematoxylin (mercury-free Haius haematoxylin; BDH) and eosin
(Eosin Technical; BDH) and examined by conventional light
microscopy.
Each section was scored independently by two assessors
according to the following criteria.
Integrity of luminal epithelium: 1, completely intact; 2, two-thirds
intact; 3, one-third intact; 4, completely absent.
Infiltration of inflammatory cells: 1, no infiltration; 2, moderate
infiltration; 3, dense infiltration.
In addition, other abnormalities, such as the degree of disten-
sion of uterine glands, the type of inflammatory cells, the presence
of glandular and stromal elements in the myometrium, as well as
the presence of oedema, were recorded.
Results
Clinical details
The 34 bitches belonged to 18 purebreeds or were crossbreeds.
Their ages ranged from one to 14 years (mean [sd], 8-4 [2-8]) and,
with the exception of two bitches which had whelped at least three
years before, they were nulliparous. None had been treated with
exogenous steroid hormones, except for a one-year-old bitch that
had received 7-5 mg oestradiol benzoate in oil after an unwanted
mating four weeks earlier during her first oestrus. The interval
from the onset of pro-oestrus ranged from one to 28 weeks (mean
[sd], 9-3 [7-8]).
Thirty of the bitches (88-2 per cent) had a vulval discharge
which was indicative of an open cervix and only three had a
closed cervix.
660 The Veterinary Record, December 12, 1998

4
co
CD CNF+
0
I-
~~~~CNF+
E
1 CNF+ |CNF+ | CNF+ CNF+ CNF+
FIG 1: Serotypes of E coli identified from
swabs obtained from the uterine lumen of 0
bitches with pyometra. K+ and K- identify
the presence or absence of surface or enve- 02 04 06 07 022 022 025 032 032 045 045 075 088 088 0147 0147
lope antigen. CNF+ identifies the presence of K+ K- K+ K+ K+ K- K- K+ K- K+ K+ K- K+ K- K- K+
cytotoxin necrotising factor Serotypes of E coli identified

Bacterial isolates
Bacteria were cultured from 28 (82-4 per cent) of the 34 cases.
Of the six that were sterile, one had a closed pyometra. E coli was
isolated from 27 (96.3 per cent) of the 28 positive cases, including
two with a closed pyometra; of these, 25 (92.5 per cent) had pure
cultures of E coli and two had a mixed infection with Klebsiella
species. The one case without E coli was infected with
Staphylococcus intermedius and Enterococcusfaecalis.
Sixteen different E coli serotypes were identified from 24 cases
(in three cases typing was not possible) (Fig 1). The most fre-
quently isolated were 032K+ (four cases) and 04K- (three cases);
022K-CNF+, 032K- and 0147K+ were each isolated from two
cases, and the other serotypes were isolated from one case each.
Cytotoxin necrotising factors (cNFs) were identified in seven of
the 16 serotypes, 02, 04, 06, 022, 025, 045 and 075 which, except
for serotypes 06 and 045, were also haemolytic.
Because of the small numbers of cases from which particular
strains of E coli were isolated, it is impossible to correlate them
with the clinical signs and their duration. In the case of 032K+,
which was isolated from four cases, the duration of the clinical
signs ranged from one to seven days.
Histopathological changes
There was considerable variation in the histopathological
changes observed in different bitches, but the correlation between
the two assessors' scoring of the lesions was high. Of the 34 tissue
sections examined, discrepancies between the individual scores
occurred in four of them and then the difference was not more
than one score. The descriptions of the glandular abnormalities,
and the type and extent of the inflammatory responses were con-
sistent for the two assessors. Tissue damage was considered to be
severe when the integrity of the luminal epithelium was complete-
ly destroyed, and there was extensive infiltration of inflammatory
cells into all layers of the uterine wall; such damage was identi-
fied in eight of the 34 cases. The classically observed greatly dis-
tended cystic uterine glands were absent in the sections from eight
of the bitches, despite the presence of a substantial inflammatory
response. In the glands and gland ducts that were distended there
were inflammatory cells or deposits of fibrin. The inflammatory
cells here, and in the tissues, were generally a mixture of mono-
nuclear cells and neutrophils.
A characteristic finding in the luminal epithelial cells when
they were intact was 'ballooning' of the cytoplasm which gave
them an irregular appearance and caused the nuclei to be dis-
placed towards the cell surface; some of these cells were also
enlarged with 'foamy' or vacuolated cytoplasm. The epithelium in
such sections was pseudostratified in appearance.
In 13 of the 34 sections, varying degrees of adenomyosis were
present.
Correlation between E coli serotypes and histopathological changes
The scores for epithelial integrity and the degree of infiltration
by inflammatory cells were compared for each serotype of E coli
isolated (Fig 2). They show that strains 04K- and 07K+ complete-
ly destroyed the integrity of the epithelium, whereas strains 088K+
and K- produced the least effect in terms of both measures of tis-
sue response. When the histopathological changes are compared
with the presence or absence of CNF1, then the mean score for the
integrity of the luminal epithelium was 3-6 for the positive group,
compared with 2-6 for the negative group; too few bitches were
involved for statistical tests. For the infiltration of inflammatory
cells, the mean scores were 2-75 for the positive group and 2-25
for the negative group.
Discussion
One of the original aims of this study was to correlate the clini-
cal signs with the bacterial flora and with the severity of the
endometrial lesions. Unfortunately, the diversity of the bacterial
isolates, particularly the E coli serotypes, and the differences in

CNF+
4-0 - * Luminal epithelial integrity
CNF+ * Cellular inflitration
3.5 - haemotytic
CNF+ CNF+
a)
CD 3.0 -
haemolytic haemolytic
CD
E
cX 2-5-
@
U)

o3
U)
w _._
0 1.5
FIG 2: Severity of tissue damage and cellular 0
C')
infiltration of the uterine wall in bitches with 0*5

pyometra in relation to the serotype of E coli

isolated. K+ and K- identify the presence or 0

absence of surface or envelope antigen. CNF+ 04 06 07 032 032 045 045 075 088 088 0147 0147
identifies the presence of cytotoxin necrotis- K- K+ K+ K+ K- K+ K+ K- K+ K- K- K+
ing factor Serotypes of E coli identified
(,3L
The Veterinary Record, December 12, 1998
the clinical approaches adopted by the individual veterinary sur-
geons, prevented such an evaluation. A logical sequel to this study
would involve a much larger sample of cases, preferably from a
single source.
In six of the 34 bitches (17.6 per cent) no bacteria could be iso-
lated from the uterine lumen, a similar proportion to that recorded
by Osbaldiston (1978) (15 per cent). It has been suggested that in
such cases the bacteria that were initially involved in the patho-
genesis of the disease may have been killed by uterine defence
mechanisms (De Vita 1952), and it is also possible that the antimi-
crobial therapy that was used in most of the cases may have been
responsible for the sterile environment.
The histological lesions observed were comparable to those
described by Dow (1958, 1959b), which included a variable
degree of cystic endometrial hyperplasia and infiltration of
inflammatory cells, depending on the severity of the disease. In
this study, eight of 29 bitches (27.5 per cent) had clinical signs of
pyometra without cystic endometrial hyperplasia. Its absence,
however, is not an unusual finding in experimental studies. For
example, Nomura and others (1988) induced pyometra in the
absence of cystic endometrial hyperplasia by the inoculation of 1
ml (5 x 107) of E coli 025 (H-K+) directly into the uterus of bitch-
es late in the luteal phase (30 to 40 days) of the oestrous cycle.
Most of the bitches which were inoculated between days 1 and 20
developed lesions of cystic endometrial hyperplasia as well. A
similar difference in response, related to the stage of the oestrous
cycle at which bitches were inoculated, has been reported by
Kang and others (1995). They found that when E coli, isolated
from clinical cases of pyometra, was inoculated into the uterus of
bitches in metoestrus, 67 per cent developed the lesions of cystic
endometrial hyperplasia, a much higher proportion than when
they were inoculated at other stages of the cycle.
Comparison between these experimental studies and sponta-
neously occurring disease may be misleading because in some
cases the cervix was ligated after the inoculation of a large num-
ber of bacteria. However, it is possible to hypothesise that in the
bitches without lesions of cystic endometrial hyperplasia, but
from which bacterial were isolated, either the infection occurred at
a stage other than early metoestrus, or the size and the nature of
the bacterial inoculum was different. Unfortunately, with only
eight cases, this hypothesis cannot be confirmed from the present
study.
E coli was isolated from 27 (79.4 per cent) of the 34 cases, and
25 of them (92.6 per cent) were in pure culture, a proportion com-
parable to the 85 per cent reported by Sandholm and others
(1975). The strains of E coli most frequently isolated, in descend-
ing order of frequency, were: 032:K+; 04:K- (CNF+); 0.22:K+
(CNF+); 03:K-; and 0147:K+, whereas Sandholm and others
(1975) found 04, 06, 02 and 075, and Nomura and others (1988)
used 025 (H-K+) previously isolated from a case of pyometra to
induce the disease experimentally. Ten of the strains had not pre-
viously been reported from cases of pyometra and, of these,
022:K+ and 025:K- also produce cytotoxic necrotising factor
(CNFI). This toxin was first isolated from strains of E coli recov-
ered from babies with enteritis; it induced necrotising lesions in
rabbits and morphological changes in HeLa and Vero cell lines.
E coli producing CNF1 has been isolated from cases of septicaemia
and urinary infections in cats and dogs (Prada and others 1991)
and in septicaemia, in the urine, vagina and faeces of dogs
(Oswald and others 1991). The significance of the presence of
CNFI is difficult to interpret at this stage, although there was some
evidence that the CNF1-positive strains were associated with
greater disruption and destruction of the luminal epithelium. It has
been suggested that, when they are present, the cytotoxin necrotis-
ing factors probably play a significant role in the pathogenesis of
the diseases with which they are associated (Prada and others
1991), which would include cystic endometrial hyperplasial
pyometra in the bitch.
Acknowledgements. - The authors thank Dr J. B. A. Smyth for
advice on interpreting the histopathological changes, Mr I.
McLaren and Mrs J. Graham for expert technical assistance, and
the veterinary practices who kindly provided material.
661
References
CAPRIOLI, A. (1983) Infection and Immunity 39, 1300
DeVITA, J. (1952) Canine Surgery, 3rd edn. Eds J. V. Lacroix, P. Hoskins.
Evanston, American Veterinary Publications Inc. p 478
DOW, C. (1958) Veterinary, Record 70, 1102
DOW, C. (1 959a) Journal of Pathology and Bacteriology 78, 267
DOW, C. (1 959b) Journal of Comparative Pathology 69, 237
GRINDLAY, M., RENTON, J. P. & RAMSAY, D. H. (1973) Research in
Veterinary Science 14, 75
HARDY, R. M. & OSBORNE, C. A. (1974) Journal of the American Animal
Hospitals Association 10, 245
KALMANSON, G. M., HARDWICK, H. J., TRUCK, M. & GUZE, L. B. (1975) The
Lancet 1, 134
KANG, B. K., PARK, 1. C. & PARK, N. Y. (1995) Korean Journal of Veterinary
Clinical Medicine 12, 31
McCABE, W. R., CARLING, P. C., BRUINS, S. & GREELY, A. (1975) Journal of
Infectious Disease 131, 6
NOMURA, K. (1984) Journal of the Japanese Veterinary Medical Association
37, 83
NOMURA, K., FUNABASHI, H. & SHIMADA, Y. (1988) Journal of the Japanese
Veterinary Medical Association 41, 95
ORSKOV, I., ORSKOV, F., JANN, B. & JANN, K. (1977) Bacteriological Reviews
41, 667
OSBALDISTON, G. W. (1978) Journal of the American Aninmal Hospitals
Association 14, 363
OSWALD, E., DeRYCKE, J., LINTERMANS, P., VAN MUYLEM, K., MAINIL,
J., DAUBE, G. & POHL, P. (199 1) Journial of Clinical Microbiology 22, 2522
PRADA, J., BALJER, G., DeRYCKE, J., STEINRUCK, H., ZIMMERMAN, S. &
BENTIN, L. (1991) Veterinary Microbiology 29, 59
SANDHOLM, M., VASENIUS, H. & KIVISTO, A-K. (1975) Journal of the
American Veterinary Medical Association 167, 1006
TEUNISSEN, G. H. B. (1952) Acta Endocrinologica 9,407
Abstract
Biochemical effects of removing
heartworms from dogs
HEARTWORMS were removed surgically from 51 dogs with
vena caval syndrome, and blood samples were taken before and
10 days after the operation. Haematological and biochemical mea-
surements showed that in the 45 dogs that survived after surgery
the red cell count and total protein, albumin, calcium, sodium and
total cholesterol concentrations increased, and total bilirubin,
ammonia, blood urea nitrogen, creatinine, uric acid and potassium
concentrations decreased. The activities of creatine kinase, lactate
dehydrogenase and amylase decreased, whereas the activities of
alkaline phosphatase and gamma-glutamyltransferase increased.
The changes in the dogs which did not survive were similar.
These changes indicate that after the worms were removed the
hepatic and renal functions of the dogs improved rapidly, presum-
ably because their general and pulmonary circulation was
improved.
KITAGAWA, H., KITOH, K., OHBA, Y., KUWAHARA, Y., IWASAKI, T. &
SASAKI, Y. (1998) Journal of the American Veterinary Medical
Association 213, 1134
The Veterinary Formulary
The fourth edition of The Veterinary Formulary, published by the
Pharmaceutical Press in association with the British Veterinary
Association, has been extensively revised and is now available.
The formulary is available, price £59.95, overseas £65, UK mem-
bers £49.95, overseas members £55 from the Pharmaceutical
Press, PO Box 151, Wallingford, Oxon, OX1O 8QU, UK, telephone
01491 824486, fax 01491 826090, email rpsgb@cabi.org.
(Access/Visa/ Mastercard/Eurocard accepted.)