Accepted Manuscript

Associations between intrauterine bacterial infection, reproductive tract inflammation

and reproductive performance in pasture-based dairy cows
Melvin de Boer, Bryce M. Buddle, Cord Heuer, Hassan Hussein, Tao Zheng, Stephen
J. LeBlanc, Scott McDougall
PII:

S0093-691X(15)00070-9

DOI:

10.1016/j.theriogenology.2015.01.032

Reference:

THE 13084

To appear in:

Theriogenology

Received Date: 22 July 2014

Revised Date:

28 January 2015

Accepted Date: 29 January 2015

Please cite this article as: de Boer M, Buddle BM, Heuer C, Hussein H, Zheng T, LeBlanc SJ,
McDougall S, Associations between intrauterine bacterial infection, reproductive tract inflammation
and reproductive performance in pasture-based dairy cows, Theriogenology (2015), doi: 10.1016/
j.theriogenology.2015.01.032.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Associations between intrauterine bacterial infection, reproductive tract

inflammation and reproductive performance in pasture-based dairy cows

Melvin de Boer a,b,*, Bryce M. Buddle c, Cord Heuer b, Hassan Hussein d, Tao Zheng c, Stephen J. LeBlanc e,

Scott McDougall a

RI
PT

SC

5
a

Cognosco, Anexa Animal Health, PO Box 21, Morrinsville 3300, New Zealand

Epicentre, Institute of Veterinary, Animal and Biomedical Sciences, Massey University, Palmerston

North 4442, New Zealand

M
AN
U

AgResearch, Hopkirk Research Institute, Grasslands Research Centre, Private Bag 11008, Palmerston

North 4442, New Zealand

11

Pokeno 2471, New Zealand

12

Department of Population Medicine, Ontario Veterinary College, University of Guelph, Guelph, Ontario

13

N1G 2W1, Canada

EP

14

TE
D

10

* Corresponding author. Tel.: +64 7 889 5159; fax: +64 7 889 3681.

16

E-mail address: mdeboer@anexa.co.nz (M.W. de Boer).

17

AC
C

15

ACCEPTED MANUSCRIPT

Abstract

19

Reproductive tract bacterial infections, particularly those caused by Escherichia coli and Trueperella

20

pyogenes can have a negative impact on reproductive performance. It has been hypothesized that

21

presence of E. coli early postpartum may increase the risk of isolation of T. pyogenes later postpartum.

22

The objective of the current study was to examine associations between intrauterine bacterial infections

23

with E. coli and T. pyogenes and any bacterial growth (irrespective of bacterial species), purulent vaginal

24

discharge (PVD), cytological evidence of endometritis (an increased proportion of polymorphonuclear

25

cells (PMN)) and reproductive performance. Dairy cows (n = 272) from six herds were examined at Days

26

0 (median: 2 d in milk), 21 and 42 postpartum. From each cow two intrauterine samples were collected

27

via triple guarded cytobrush at Days 0 and 21. The first cytobrush was used for bacteriological culture.

28

Escherichia coli and T. pyogenes were isolated by culture, and E. coli isolates were assigned to 1 of 4

29

phylogenetic groups using a two-step triplex polymerase chain reaction (PCR). In addition, T. pyogenes

30

was confirmed by PCR. The second cytobrush was used to prepare a cytology slide. Nucleated cells (n =

31

200) were categorized as epithelial cells, PMN or macrophages. Cows were also assessed for body

32

condition score, PVD score, presence of a corpus luteum and pregnancy. Statistical analysis was

33

performed using multivariable models. There was no association between presence of E. coli at Day 0

34

and probability of isolation of T. pyogenes 3 wk later, however E. coli positive cows at Day 0 were more

35

likely to be diagnosed with E. coli at Day 21 (RR = 2.0, P < 0.01). E. coli at Day 0 or T. pyogenes at Day 21

36

increased the risk of PVD diagnosis 3 wk later (RR = 1.9; P = 0.04 and RR = 3.0; P = 0.05, respectively).

37

Cows with any bacterial growth at Day 21, irrespective of species, were less likely to conceive within 3

38

wk after the start of the seasonal breeding program (RR = 0.8; P = 0.05). Interestingly, cows with 25%

39

PMN at Day 0 had shorter time to pregnancy (HR = 1.32; P = 0.05). Intrauterine bacterial infection may

40

impair reproductive performance but presence of E. coli was not associated with isolation of T. pyogenes

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

18

ACCEPTED MANUSCRIPT

41

3 wk later. Increased endometrial flux of PMN in cows early postpartum may be a physiological process

42

and improve reproductive performance.

RI
PT

43
44

Keywords: Escherichia coli, Trueperella pyogenes, purulent vaginal discharge, endometritis,

45

polymorphonuclear cells, conception

SC

46
1. Introduction

48

Postpartum, most cows have bacteria present in the uterus. However, a majority of cows undergo self-

49

resolution with time [1]. A wide diversity of bacterial species are present in the uterus of cows, but

50

differences in isolated bacterial species occur between healthy cows and those with reproductive tract

51

disease [2]. Bacteria which are considered potentially pathogenic to uterine tissue include Escherichia

52

coli and Trueperella pyogenes [3]. Risk factors for uterine disease include low body condition score (BCS)

53

at calving, hyperketonemia, and peripartum disease [1, 4].

54

Endometritis may be defined as cytological evidence of inflammation of endometrial tissue which is

55

associated with reduced reproductive performance [5]. Recent studies have demonstrated that purulent

56

vaginal discharge (PVD) may be associated with, but is not identical to endometritis [6]. Therefore, a

57

diagnosis of PVD is applied to cows with grossly evident purulent material in the vagina, whether or not

58

there is evidence of endometritis as defined by cytology [5]. Purulent vaginal discharge can be

59

diagnosed by retrieval of purulent material from the vagina by placement of a gloved hand [7] or

60

MetricheckTM device [1] into the vagina, or visualization by vaginoscopy [8]. The Metricheck device, a

61

soft rubber hemisphere connected to a stainless steel rod, is inserted into the vagina. Vaginal discharge

62

is evaluated after retracting the device caudally [1]. Endometritis may also be diagnosed by endometrial

AC
C

EP

TE
D

M
AN
U

47

ACCEPTED MANUSCRIPT

biopsy [9]. Uterine cytology following sample collection by cytobrush or uterine lavage are now

64

frequently used, at least in the research environment [5]. The cytobrush technique collects cells for

65

cytology by rolling a small brush against the endometrium [10].

66

The same sample collected to define the degree of uterine inflammation may also be used for

67

bacteriology [2, 9, 11]. Isolation of some species of bacteria from the uterus early postpartum have been

68

associated with reduced dominant follicle growth rate, reduced estradiol concentrations and a smaller

69

corpus luteum following ovulation [12]. However, not all studies have found associations between

70

isolation of bacteria and reproductive outcomes [13]. Specific strains or virulence factors of E. coli seem

71

to be more pathogenic than others [14, 15], and intrauterine presence of a specific virulence factor for

72

E. coli (fimH) was associated with reduced pregnancy rate [15]. Cows infected with T. pyogenes took

73

longer to conceive compared to those without T. pyogenes [16]. The presence of T. pyogenes has also

74

been associated with an increased influx of inflammatory cells in uterine tissue [9] and a higher risk of

75

PVD [3].

76

The prevalence of E. coli decreases and that of T. pyogenes increases with time postpartum [11, 12].

77

Additionally, a small study in which approximately half of the cows had retained foetal membranes

78

(RFM) found an association between E. coli within 2 d postpartum and T. pyogenes at 14 DIM (days in

79

milk) [17].

80

The primary objective of this study was to test the hypothesis that the presence of E. coli within the first

81

wk postpartum increases the risk of subsequent isolation of T. pyogenes. Secondary objectives were to

82

assess associations between early isolation of E. coli, T. pyogenes or any bacteria (i.e. isolation of any

83

bacterial species irrespective of type) and subsequent isolation of these bacteria, reproductive tract

84

inflammation measured by PVD or polymorphonuclear cells (PMN) in endometrial cytology, and

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

63

ACCEPTED MANUSCRIPT

reproductive performance. The final objectives were to assess associations between early reproductive

86

tract inflammation and subsequent inflammation, and reproductive performance.

87

2. Materials and Methods

88

Animal ethics approval was obtained prior to study commencement (AgResearch Animal Ethics

89

Committee; Ruakura; Hamilton; New Zealand; application number: 12395).

90

2.1. Herds and cows

91

Dairy cows (n = 272) from six commercial herds in the Waikato region of New Zealand were enrolled

92

between 18 July and 12 August 2011 (Table 1, Fig. 1). Data collection ended 7 March 2012, the date of

93

the last pregnancy diagnosis. The enrolment of herds was on a convenience basis, with herds being

94

recruited based on willingness to participate, agreeing to the study protocol and allowing access to

95

herd- and cow-level records. Cows were managed in spring-calving, predominantly pasture-fed herds. At

96

each enrolment day (Day 0), herd owners were asked to present multiparous cows calved between 1

97

and 4 d for veterinary examination. Cows with peripartum disease (e.g. dystocia, retained foetal

98

membranes) were included, but cows were excluded if an antibiotic or anti-inflammatory treatment had

99

been administered 30 d before enrolment or if rectal temperature >39.5 oC was recorded at the initial

TE
D

M
AN
U

SC

RI
PT

85

examination. When 25 cows were available on a given enrolment day (maximum two days for each

101

herd), all cows that met the enrolment criteria were enrolled. When >25 were available, the first or

102

second cow in the row was selected for exclusion by flip of a coin. Thereafter every second cow in the

103

row was excluded from the study until in total 25 cows remained.

104

2.2. Sampling methods

105

Cows were sampled on Days 0, 21 (range: 0 d) and 42 (2 d). On Day 21, cytobrush samples were taken

106

by one of two veterinarians (MdB and SL), while all other samples and measurements, including those

AC
C

EP

100

ACCEPTED MANUSCRIPT

taken at Days 0 and 42, were taken by the one veterinarian (MdB). On Day 0, the following

108

measurements and samples were collected from each study cow: rectal temperature, uterine tone score

109

(firm vs. large/flaccid without striations), BCS on a 1 to 10 scale [18], vaginal discharge score (VDS)

110

evaluated using a Metricheck device (Simcro Tech, Hamilton, New Zealand) and two intrauterine

111

samples using cytobrushes (Ebos Group Ltd, Auckland, New Zealand). A VDS of 0 was given when no

112

mucus was visible; a score of 1, 2, 3, 4, or 5, was given when clear mucus, flecks of purulent discharge,

113

more than flecks but less than 50% purulent discharge, more than 50% purulent discharge, more than

114

50% purulent discharge plus odor, respectively, was visualized [1]. Two intrauterine samples were

115

collected per cow using a triple-guarded cytobrush as previously described [19]. The first cytobrush was

116

used for bacterial isolation and was immediately placed in transport media (Amies Agar Gel, Fort

117

Richard, Auckland, New Zealand) and stored on ice. After moving the catheter slightly forward, a second

118

sample was taken for cytology. After retraction, this brush was rolled gently along a clean microscope

119

slide. All measurements, with the exception of rectal temperature were repeated on Day 21. On Day 42,

120

BCS and VDS were recorded, and presence of a CL was determined using transrectal ultrasonography

121

with a variable frequency (4.5 to 8.5 MHz) transducer (Easi-Scan, BCF Ultrasound, Auckland, New

122

Zealand). The presence and stage of gestation was determined using transrectal ultrasonography

123

between 10-12 wk after the start of the seasonal breeding program (planned start of mating; PSM) and

124

again at approximately 6 wk after the bulls were removed from the herd.

125

On Day 42, following common practice in New Zealand [1], each study cow with a VDS 2 was treated

126

with 500 mg cephapirin i.u. (MSD Animal Health, New Zealand; Table 1). Day 42 was a median of 29 d

127

(range: 19 to 37 d) before PSM. Cows not detected in estrus by removal of tail paint 10 d before PSM

128

were treated with a combination of the Ovsynch protocol and a vaginal progesterone insert at the PSM

129

(an

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

107

ACCEPTED MANUSCRIPT

estrus treatment), followed by set time AI [20]. Following PSM cows were artificially inseminated based

131

on detection of estrus using tail paint for 4 to 8 wk, followed by a period where bulls were run with the

132

herds (Table 1).

133

2.3. Bacteriology

134

The first cytobrush was used to streak a 0.1% esculin, 5% sheep blood agar plate and a MacConkey agar

135

plate (Fort Richard, Auckland, New Zealand) on the day of sampling. The blood agar plate was placed

136

into a sealed GasPackTM container (BD, Auckland, New Zealand) with CO2-enriched atmospheres. The

137

MacConkey plate was incubated aerobically. Bacterial growth of each plate was assessed after 2 d of

138

incubation at 37 oC. Culture positive plates, irrespective of bacterial species was recorded as having

139

bacterial growth. Of the cultured bacteria, only E. coli and T. pyogenes were typed. The identity of

140

presumptive E. coli and T. pyogenes was confirmed on the basis of colony morphology, Gram stain and

141

biochemical tests [21]. Following subculture, each purified isolate was inoculated into 2 ml brain heart

142

infusion broth with 20% glycerol and stored at -20 oC for further analyses. To determine phylogenetic

143

groups A, B1, B2 and D, E. coli isolates were subjected to a two-step triplex PCR using primers targeting

144

the genes chuA and yjaA and the DNA fragment TspE4.C2 (Table 2). Confirmation of T. pyogenes isolates

145

was undertaken using PCR using primers targeting the genes plo and sodA, and the 16S-23S rDNA

146

intergenic spacer region (Table 2).

147

2.4. Cytology

148

Slides were stained with a modified Wright-Giemsa stain (Diff-Quick, New Zealand Veterinary Pathology

149

Ltd, Hamilton, New Zealand), air dried, and stored. Two hundred nucleated cells (i.e. epithelial cells,

150

PMN and macrophages; MAC) were counted at a 400x magnification by one veterinarian (MdB).

151

Percentages of PMN (PMN%) and MAC (MAC%) were calculated by dividing the number of PMN or MAC

152

from the total number of the identified nucleated cells. A subset of slides (n = 53; 10%) were randomly

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

130

ACCEPTED MANUSCRIPT

selected and evaluated twice to assess intraobserver repeatability. Intraobserver repeatability of PMN%

154

assessment was found to be high (PMN%: rc = 0.94, P < 0.001, and MAC%: rc = 0.89, P < 0.001).

155

2.5. Power statistics

156

It was anticipated that the prevalence of intrauterine E. coli within 4 d postpartum would be 25% [22],

157

and that there would be a critical difference of 20% points in prevalence of T. pyogenes between

158

exposed (E. coli positive) and non-exposed (E. coli negative), i.e., 57% T. pyogenes in cows that had E.

159

coli just after calving and 37% T. pyogenes in non-exposed (i.e. 0.25 x 57% + 0.75 x 37% = 42%) [22].

160

With a ratio of 25%:75% = 1:3 of exposed vs. non-exposed, a sample size of 220 cows was needed

161

yielding 55 exposed and 165 non-exposed (80% power, P = 0.05, 2-sided).

162

2.6. Statistical analysis

163

Data were collected in a purpose-built SQL database (Microsoft SQL Server Management Studio 2008,

164

Seattle, WA) and transferred for statistical analyses into Stata 13.1 (StataCorp, College Station, USA).

165

To asses any potential bias, cows removed from the study were compared with those not removed using

166

ANOVA and 2 statistics for continuous and categorical variables, respectively, for herd, age, breed, DIM

167

at Day 0, and presence of E. coli, T. pyogenes, PMN%, and MAC% at Days 0 and 21. Intraobserver

168

variability of the estimate of PMN percentage from the cytology slides was assessed using Lins

169

concordance correlation coefficient [23].

170

For multivariable analyses, the following predictor variables were considered: age (categorized into 3

171

and 4, 5 and 6, >6 years); breed (categorized into Friesian (>11/16th Friesian), Jersey (>11/16th Jersey)

172

and the remaining defined as crossbreed); time from the herds planned start of calving to the cows

173

actual calving date (days); DIM at Day 0; rectal temperature at Day 0; uterine tone (categorized as firm

174

or large/flaccid) at Days 0 and 21; BCS (categorized as 4: thin and >4: normal, no obese cows present)

175

at Days 0, 21 and 42; presence or absence of E. coli at Days 0 and 21; presence or absence of T.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

153

ACCEPTED MANUSCRIPT

pyogenes at Days 0 and 21; presence or absence of any bacterial growth, irrespective of species, at Days

177

0 and 21; PVD categorized as absent (i.e. VDS 1) or present (i.e. VDS > 1) at Days 0, 21 and 42; PMN%

178

at Days 0 and 21; MAC% at Days 0 and 21; presence or absence of a CL at Day 42, and detection of

179

estrus (yes or no) by PSM. Isolated E. coli were categorized into phylogenetic groups A, B1, and B2/D.

180

Phylogenetic groups B2 and D are extra-intestinal pathogenic strains with multiple virulence factors [24],

181

and due to the low proportion of group B2, this group was grouped with group D strains for analysis.

182

Dystocia was not included in the analyses, because one herd (n = 45) failed to record this and due to the

183

low incidence in the remaining herds (<1%; n = 2/208).

184

Reproductive tract infection was defined as isolation of E. coli, T. pyogenes, or any bacterial growth

185

(irrespective of species), whereas reproductive tract inflammation was assessed by the variables PVD or

186

PMN%. The following associations were assessed: uterine infection status at each sample day with

187

infection status on subsequent days, with subsequent reproductive inflammation, with anestrus

188

treatment and reproductive performance. Additional assessed associations were reproductive tract

189

inflammation at each sample day with subsequent days, anestrus treatment and reproductive

190

performance. The Days 21 and 42 and the reproduction variables listed in Fig. 2 were considered as the

191

outcome variables for multivariable analyses. Variables were added to multivariable binary logistic and

192

linear regression models using manual forward stepwise regression and retained if the P-value was

193

0.10 or if it caused a change in one of the coefficients by >20%. Akaikes information criterion was used

194

for model comparison. Model fit for multivariable binary logistic regression was determined using

195

Hosmer-Lemeshow test and influential points were evaluated by plotting the leverage of individual data

196

points, Pearsons raw and standardized residuals, and the deviance of residuals against the predicted

197

values. Model fit for multivariable linear regression was evaluated by a QQ-plot of the raw residuals, and

198

plotting the raw and standardized residuals, and the leverage of individual data points against the

199

predicted values. Herd was included as a fixed effect in each model. Interaction terms were evaluated

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

176

ACCEPTED MANUSCRIPT

where biologically plausible. Relative risk (RR) is preferred over odds ratios (OR) when the outcomes are

201

common, because OR can be a considerable overestimation of associations [25]. Poisson regression was

202

used to compute RR for binary outcomes [25].

203

Survival analysis was used to analyze the PSM to conception interval. Cows removed during the breeding

204

period were right censored on the day of removal, and those not pregnant at the end of the breeding

205

period were right censored on the day of bull removal. Independent variables as described above (P

206

0.10) were offered to Coxs proportional hazards models using manual forward stepwise regression. The

207

analysis was performed using previously published cut-points for PMN% of 5% [26], 10% and 18% [27],

208

and 25% [28]. The proportional hazards assumption was evaluated by plotting Log[-log S(t)] and the

209

Shoenfeld residuals against time, and by assessing the Shoenfeld residuals and time-dependent

210

interaction term for the covariates. Significance was declared at P 0.05, and a tendency at P 0.10.

211

3. Results

212

3.1. Descriptive statistics

213

Retained foetal membranes were diagnosed in two cows selected for enrolment, of which one did not

214

meet the enrolment criteria because of pyrexia and treatment with antimicrobials at Day 0. The second

215

cow with RFM was enrolled, but subsequently excluded following antimicrobial treatment four days

216

later. Including the enrolled cow with RFM, 16/272 cows (6%) had received antimicrobial or anti-

217

inflammatory treatment before Day 21. One cow was diagnosed with a gastrointestinal tract accident

218

and was culled, and two cows enrolled without appropriate herd identification and were lost to follow

219

up. Of the 272 enrolled study cows that were sampled at Day 0, 19 (7%) were excluded after enrolment,

220

leaving 253 cows for analysis. Additional descriptive statistics are presented in Table 1.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

200

10

ACCEPTED MANUSCRIPT

3.2. Microbiology and cytology

222

The prevalence of isolation of any bacteria, i.e. culture positive irrespective of bacterial species, was 93%

223

at Day 0 and 79% at Day 21. Overall, the prevalence of E. coli decreased from 42% at Day 0 to 24% at

224

Day 21 (Table 3). The distribution of the E. coli phylogenetic groups is presented in Table 4. At Day 0, T.

225

pyogenes was not detected in four of the six herds and the overall prevalence was low (1% of cows) but

226

increased to 9% by Day 21. The distribution of uterine bacteria varied amongst herds (Table 3).

227

The overall prevalence of PVD was 10%, 15% and 7% at Days 0, 21, and 42, respectively (Table 1). The

228

prevalence of PVD at Day 21 varied between herds (P 0.05). Of the 527 slides assessed for cytology,

229

two slides at each of Days 0 and 21 were of poor quality (i.e. low number of cells or distorted cells) and

230

were excluded from further analysis. Polymorphonuclear cells were identified in the majority of cows

231

(96% and 87% at Day 0 and Day 21, respectively). The median PMN% were 41% (range: 0-95%) and 5%

232

(0-84%) at Days 0 and 21, respectively (Fig. 3a and b). In total, 89% (range between herds: 80-94%), 79%

233

(72-84%), 68% (60-73%) and 62% (53-73%) of cows at Day 0, and 50% (42-58%), 36% (28-41%), 26% (13-

234

34%) and 19% (4-26%) of cows at Day 21 had a PMN% above 5%, 10%, 18%, and 25%, respectively.

235

Macrophages were identified in 78% and 38% cows at Days 0 and 21, respectively. The median MAC%

236

were 7% (range: 0-82%) and 0% (0-67%) at Days 0 and 21, respectively.

237

3.3. Model fit and diagnostics of multivariable analysis

238

A diagram of all associations (P 0.10) is presented in Fig. 2. The Hosmer-Lemeshow test for each binary

239

logistic regression model showed a good fit (P > 0.05). Model diagnostics found a small number of

240

outliers in four models, however removing those did not improve the diagnostics. All cows were

241

included in the final models in those circumstances, indicating that the model may be an under-

242

prediction. Only for the models with the outcome variables of anestrus (logistic regression) and PMN%

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

221

11

ACCEPTED MANUSCRIPT

at Day 21 (linear regression) two outliers in each model were removed to improve model diagnostics.

244

The proportional hazards assumption for the time to pregnancy analysis was not violated.

245

3.4. Bacteriological multivariable associations

246

Isolation of E. coli at Day 0, while accounting for breed and DIM at enrolment, was not associated with

247

the isolation of T. pyogenes at Day 21 (P = 0.53). Additionally, there were no associations (P > 0.10 for

248

each group) between the phylogenetic group of E. coli isolated at Day 0 and isolation of T. pyogenes at

249

Day 21. However, isolation of E. coli at Day 0 increased the risk of isolation of E. coli or of any bacterial

250

growth at Day 21, with 34/105 (32%) with E. coli and 25/148 (17%) without E. coli at Day 0, having E. coli

251

isolated at Day 21 (P < 0.01); and 91/105 (87%) cows with E. coli and 111/148 (75%) of cows without E.

252

coli at Day 0, having any bacterial growth at Day 21 (P = 0.03).

253

Cows with E. coli at Day 0 and those with T. pyogenes at Day 21 were more likely to be diagnosed with

254

PVD 3 wk later, with 20/105 (19%) with E. coli and 17/148 (11%) without E. coli at Day 0, being

255

diagnosed with PVD at Day 21 (P = 0.04); and 4/23 (17%) cows with T. pyogenes and 14/230 (6%) of

256

cows without T. pyogenes at Day 21, being diagnosed with PVD at Day 42 (P = 0.05). In contrast, those

257

with E. coli at Day 21 tended to be less likely to be diagnosed with PVD at 42, with 1/59 (2%) with E. coli

258

and 17/194 (9%) without E. coli at Day 21, being diagnosed with PVD at Day 42 (P = 0.10). Associations

259

between bacteriological variables and reproduction outcomes are presented in Fig. 2.

260

3.5. Associations with reproductive tract inflammation

261

A standard deviation increase over the mean of PMN% at Days 0 or 21 increased the probability of being

262

diagnosed with PVD 3 wk later from 13 to 21% (P < 0.01), and from 7 to 9% (P = 0.07), respectively.

263

Polymorphonuclear cell percentage at Day 0 was significantly associated with PMN% at Day 21 as

264

calculated by linear regression (P < 0.01; Fig. 2). Associations between inflammatory variables and

265

reproduction outcomes are presented in Fig. 2.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

243

12

ACCEPTED MANUSCRIPT

3.6. Variables affecting time from the start of breeding program to conception

267

Cows with E. coli isolated at Day 21 tended to conceive later than those without (hazard ratio = 0.8 (95%

268

CI: 0.5-1.0); P = 0.08, Fig. 4). Conception occurred earlier in cows with 25% PMN at Day 0 than those

269

with <25% PMN (hazard ratio = 1.3 (95% CI: 1.0-1.7); P = 0.05, Fig. 5).

270

4. Discussion

271

The primary objective of this study was to assess whether isolation of E. coli from the bovine uterus

272

within the first wk postpartum would predispose to isolation of T. pyogenes 3 wk later. Additionally,

273

associations between early isolation of E. coli, T. pyogenes or any bacteria (i.e. isolation of any bacteria

274

irrespective of species) and subsequent bacterial isolation, reproductive tract inflammation, and

275

reproductive performance were assessed. Finally, effects of early postpartum reproductive tract

276

inflammation and subsequent inflammation, and reproductive performance were evaluated.

277

Primiparous cows were excluded from this study. Differences in reproductive tract inflammation

278

between primiparous and multiparous cows exist, and therefore it has been suggested that these parity

279

groups should be assessed separately [29]. Further, this study involved multiple herds, whereas others

280

did not. It was expected that associations between uterine pathogens would vary between herds and

281

that the external validity of the study would be improved by enrolling multiple herds. However, no

282

interactions were observed between those associations and herd, possibly due to considerable variation

283

in natural pathogen challenge (i.e. prevalence within herd; Table 3) or limited statistical power.

284

Nevertheless, many previous studies examining uterine bacteriology in dairy cows have used only one

285

herd [3, 13, 29], hence extrapolation of these results to dairy cows in general must be undertaken with

286

caution.

287

The decreasing prevalence of E. coli and increasing prevalence of T. pyogenes over time resulted in the

288

suggestion that E. coli early postpartum predisposes cows to infection with T. pyogenes [12]. Similar to

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

266

13

ACCEPTED MANUSCRIPT

others using traditional culture techniques [11], such an association was not found in this study.

290

Additionally, no significant association was found in the current study between any of the phylogenetic

291

groups of E. coli and subsequent isolation of T. pyogenes. However, it has been reported that particular

292

genotypes within E. coli phylogenetic group B1 isolated early postpartum were significantly associated

293

with subsequent isolation of T. pyogenes [14]. Even though T. pyogenes affected few cows, its presence

294

seemed to be of clinical importance and was associated with reproductive tract inflammation in our

295

study and impaired reproduction as described by others [11, 16]. Besides decreasing E. coli and

296

increasing T. pyogenes proportions over time, little evidence, including the present data, is currently

297

available supporting the notion that presence of E. coli predisposes to T. pyogenes infection. However,

298

more cows may be needed in future natural exposure studies to prove this hypothesis, as the power to

299

test the primary hypothesis of this study was only moderate. Post hoc power calculation showed that

300

the sample size would have to have been 3-fold greater to define the differences observed in the

301

current study as significant.

302

A study using culture independent techniques, i.e. genomic tests, recently reported that E. coli was less

303

prevalent compared to Fusobacterium necrophorum [15]. That same study found that presence of F.

304

necrophorum, but not E. coli, increased the odds of subsequent detection of T. pyogenes. Similarly,

305

another study also failed to find an association between E. coli and T. pyogenes using culture

306

independent techniques, but found that F. necrophorum was highly prevalent in metritic cows, but not

307

detected in healthy cows [2]. In our study, assessing the association between E. coli and T. pyogenes was

308

the main objective and thus these were the only bacterial species typed. Therefore, it is unknown

309

whether F. necrophorum predisposes T. pyogenes infection in pasture-based cows and further research

310

is required. Sensitivity of culture independent bacteriology techniques seems to be higher than culture-

311

based techniques, although specificity may be lower because of the possibility of detecting related or

312

non-viable bacteria [30]. Accuracies of culture dependent versus independent techniques have not yet

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

289

14

ACCEPTED MANUSCRIPT

been compared for intrauterine bacteriological tests [31]. The current study used culture-based

314

techniques and may have underestimated the true prevalence of E. coli and T. pyogenes. Further studies

315

are required to compare techniques. Due to the lack of a gold standard, such studies may use

316

reproductive performance outcomes for comparing and validating tests [5].

317

The presence of E. coli may truly not actually predispose to T. pyogenes infections, and anaerobic

318

pathogens, such as F. necrophorum may be more clinically important early postpartum. Other

319

underlying etiologies, such as infection with bovine herpes virus type 4 (BoHV-4) may be associated with

320

increased risk of T. pyogenes. A relationship between BoHV-4, E. coli, and T. pyogenes uterine infections

321

has been suggested by Sheldon et al. [32], with E. coli and T. pyogenes promoting the replication of

322

BoHV-4, resulting in greater pathology to endometrial cells and more inflammation. Antibodies against

323

BoHV-4 are present in New Zealand dairy cows [33], but contribution of this virus to reproductive

324

infection was not the objective of current study, hence further research is needed.

325

The association between intrauterine bacterial growth and reproductive tract inflammation has been a

326

subject of interest for many years and controversy still exists [9, 13]. Similar to Deguillaume et al. [34],

327

this study did not find an association between bacterial intrauterine presence and MAC%. The PMN% at

328

Day 21 in this study was numerically higher compared to others sampling cows up to 29 DIM [13, 29].

329

This study found no association between the bacteriological variables at Day 0 and increased PMN% 3

330

wk later. This stands in contrast to another study, where cows without PVD and with T. pyogenes

331

isolated at 10 1 DIM had increased odds of PMN (>18%) 2 wk later [29]. That study, similar to the

332

current study, did not find an association between E. coli and PMN%. McDougall et al. [13] reported

333

little agreement between one of five major bacteria and PMN% at 29 DIM in pasture-based cows.

334

Besides exclusion of PVD, other differences in cytological outcomes like variations in sampling protocol,

335

number of cells counted, protocol of cytology slide assessment, and examined populations complicate

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

313

15

ACCEPTED MANUSCRIPT

comparisons between studies [5]. Based on data from our study and others [13], it appeared that

337

bacteria isolated from the uterus of pasture-based cows were not associated with PMN%. It is possible

338

that cows which calve on pasture may be contaminated with different species of bacteria compared to

339

those calving in a barn; potentially explaining differences amongst studies undertaken in different

340

environments.

341

Cows identified with E. coli at Day 0 (attributable fraction = 0.40) or T. pyogenes at Day 21 (attributable

342

fraction = 0.65) were at increased risk of being diagnosed with PVD 3 wk later, confirming findings of

343

others [3, 11]. Specific bacterial genotypes or phenotypes may be a more important cause of

344

reproductive tract inflammation than the presence of the bacterial species [15]. The current study did

345

not find any associations between E. coli phylogenetic groups and reproductive tract inflammation. In

346

contrast, one study described an association between some genotypes within the phylogenetic groups A

347

and B1 and PVD, but failed to report if the association was significant or not [14]. Interestingly, PVD at

348

Day 42 tended to be diagnosed less often when E. coli was isolated at Day 21, however caution needs to

349

be taken interpreting this non-plausible result as causal, because of the low proportion of cows (2%)

350

with E. coli at Day 21 and being diagnosed with PVD at Day 42. Hence, it may be a type 1 error.

351

Escherichia coli endotoxin lipopolysaccharide disrupts ovarian function by interfering with the

352

hypothalamus-pituitary axis [12], and switches endometrial production of prostaglandin F2 (PGF2) to

353

prostaglandin E2 (PGE2) [35]. PGE2 has been shown to suppress some immune functions [36], but it is

354

unclear if this hormone affects leucocyte extravasation. Thus, questions remain to why, unexpectedly, E.

355

coli isolated at Day 21 decreased the probability of PVD diagnosis and increased the probability of the

356

presence of a CL at Day 42, yet was associated with longer time to pregnancy.

357

Isolation at Day 21 of E. coli or any bacteria irrespective of species was associated with increased time to

358

pregnancy and a reduced probability of pregnancy within 3 wk from PSM, respectively (Fig. 2 and 4). In

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

336

16

ACCEPTED MANUSCRIPT

contrast, no associations were found between isolation of one or more of the five major bacterial

360

species (i.e. bacterial species likely to be pathogenic for the endometrium, including E. coli and T.

361

pyogenes) at 29 DIM and reproductive outcomes in another study in New Zealand [13]. Whereas, T.

362

pyogenes at 26 days postpartum reduced probability of pregnancy by 119 days postpartum [16].

363

Similarly, E. coli possessing the virulence factor fimH was associated with increased time to pregnancy

364

[15]. A recently reported reduction in reproductive tract disease by prepartum vaccination against E. coli

365

and other putative uterine pathogens [37] supports the notion that reduction of uterine pathogen load

366

or its effects in the early postpartum period is a relevant goal.

367

New Zealand dairy cows are commonly tested for PVD about a month before PSM. In a large study,

368

those diagnosed with PVD but not treated were at higher risk of reproductive failure [1]. Therefore,

369

cows with PVD on Day 42 (approximately one month before PSM) were in the present study treated at

370

that time, thereby diminishing a potential negative affect of PVD on reproductive performance, hence it

371

was not evident. Many studies have reported impaired reproductive performance in cows with

372

increased PMN% at various time points [19, 27]. The protective effect of PMN% at Day 0 on time to

373

conception from PSM appears to be a novel finding from this study (Fig. 5). In addition, Gilbert et al. [38]

374

reported that the presence of bacteria and the risk of endometritis as determined by PMN% was

375

reduced by the influx of PMN early postpartum. It appears from that data and our study that an influx of

376

PMN may be part of the physiological process in early postpartum cows and required to achieve optimal

377

reproductive performance. Thus, early postpartum treatment with anti-inflammatory drugs in non-

378

systemically ill cows may potentially be contraindicated. Recently, no benefit was found following early

379

postpartum treatment with a non-steroidal anti-inflammatory drug [39].

380

In conclusion, this study demonstrated no association between culturing E. coli from the uterus at Day 0

381

and T. pyogenes 3 wk later, while presence of E. coli did increase the probability of subsequent re-

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

359

17

ACCEPTED MANUSCRIPT

isolation of the same bacterial species. Although the presence of E. coli, T. pyogenes, or any bacterial

383

species did not influence PMN% 3 wk later, the clinical importance of detection of E. coli at Day 0 and T.

384

pyogenes at Day 21 was the association with the subsequent diagnosis of PVD. Additionally, cows with

385

intrauterine isolation of E. coli or culture of any bacteria irrespective of species had poorer reproductive

386

performance. Interestingly, cows with PMN 25% at Day 0 were pregnant sooner compared to those

387

below this cut-point, indicating that an early influx of PMN may be part of the cows physiological

388

response of uterine involution. Conversely, PMN% at day 21 was negatively associated with pregnancy

389

with 3 wk from PSM, indicating inflammation at that stage is detrimental.

390

Acknowledgements

391

The cooperation of the farmers and their staff is gratefully acknowledged. The authors would like to

392

thank Sbastien Hudault in particular for on farm and laboratory technical support, and Joanne

393

Niethammer, Cathy Yanez, Elizabeth Blythe, Amanda Hallett and Peta Rossiter for additional technical

394

support. Andy Taylor helped with database creation and support and Jaimie Hunnam assisted with

395

manuscript revision. This study was funded by Cognosco, Anexa Animal Health, Morrinsville and

396

AgResearch, Hopkirk Research Institute, Palmerston North.

397

References

398

[1] McDougall S, Macaulay R, Compton C. Association between endometritis diagnosis using a novel

399

intravaginal device and reproductive performance in dairy cattle. Anim Reprod Sci. 2007;99:9-23.

400

[2] Santos TMA, Bicalho RC. Diversity and succession of bacterial communities in the uterine fluid of

401

postpartum metritic, endometritic and healthy dairy cows. PLoS ONE. 2012;7:e53048.

402

[3] Williams EJ, Fischer DP, Pfeiffer DU, England GCW, Noakes DE, Dobson H, et al. Clinical evaluation of

403

postpartum vaginal mucus reflects uterine bacterial infection and the immune response in cattle.

404

Theriogenology. 2005;63:102-17.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

382

18

ACCEPTED MANUSCRIPT

[4] Dubuc J, Duffield TF, Leslie KE, Walton JS, LeBlanc SJ. Risk factors for postpartum uterine diseases in

406

dairy cows. J Dairy Sci. 2010;93:5764-71.

407

[5] De Boer MW, LeBlanc SJ, Dubuc J, Meier S, Heuwieser W, Arlt S, et al. Invited review: Systematic

408

review of diagnostic tests for reproductive-tract infection and inflammation in dairy cows. J Dairy Sci.

409

2014;97:3983-99.

410

[6] Dubuc J, Duffield TF, Leslie KE, Walton JS, LeBlanc SJ. Definitions and diagnosis of postpartum

411

endometritis in dairy cows. J Dairy Sci. 2010;93:5225-33.

412

[7] Plntzke J, Madoz LV, De la Sota RL, Heuwieser W, Drillich M. Prevalence of clinical endometritis and

413

its impact on reproductive performance in grazing dairy cattle in Argentina. Reprod Domest Anim.

414

2011;46:520-6.

415

[8] Leutert C, von Kruger X, Plntzke J, Heuwieser W. Evaluation of vaginoscopy for the diagnosis of

416

clinical endometritis in dairy cows. J Dairy Sci. 2012;95:206-12.

417

[9] Bonnett BN, Martin SW, Gannon VPJ, Miller RB, Etherington WG. Endometrial biopsy in Holstein-

418

Friesian dairy cows. 3. Bacteriological analysis and correlations with histological-findings. Can J Vet Res.

419

1991;55:168-73.

420

[10] Kasimanickam R, Duffield TF, Foster RA, Gartley CJ, Leslie KE, Walton JS, et al. A comparison of the

421

cytobrush and uterine lavage techniques to evaluate endometrial cytology in clinically normal

422

postpartum dairy cows. Can Vet J. 2005;46:255-9.

423

[11] Werner A, Suthar V, Plntzke J, Heuwieser W. Relationship between bacteriological findings in the

424

second and fourth weeks postpartum and uterine infection in dairy cows considering bacteriological

425

results. J Dairy Sci. 2012;95:7105-14.

426

[12] Williams EJ, Fischer DP, Noakes DE, England GCW, Rycroft A, Dobson H, et al. The relationship

427

between uterine pathogen growth density and ovarian function in the postpartum dairy cow.

428

Theriogenology. 2007;68:549-59.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

405

19

ACCEPTED MANUSCRIPT

[13] McDougall S, Hussein H, Aberdein D, Buckle K, Roche J, Burke C, et al. Relationships between

430

cytology, bacteriology and vaginal discharge scores and reproductive performance in dairy cattle.

431

Theriogenology. 2011;76:229-40.

432

[14] Sheldon IM, Rycroft AN, Dogan B, Craven M, Bromfield JJ, Chandler A, et al. Specific strains of

433

Escherichia coli are pathogenic for the endometrium of cattle and cause pelvic inflammatory disease in

434

cattle and mice. PLoS ONE. 2010;5:e9192.

435

[15] Bicalho MLS, Machado VS, Oikonomou G, Gilbert RO, Bicalho RC. Association between virulence

436

factors of Escherichia coli, Fusobacterium necrophorum, and Arcanobacterium pyogenes and uterine

437

diseases of dairy cows. Vet Microbiol. 2012;157:125-31.

438

[16] Bonnett BN, Martin SW, Meek AH. Associations of clinical findings, bacteriological and histological

439

results of endometrial biopsy with reproductive performance of postpartum dairy cows. Prev Vet Med.

440

1993;15:205-20.

441

[17] Dohmen MJW, Joop K, Sturk A, Bols PEJ, Lohuis J. Relationship between intra-uterine bacterial

442

contamination, endotoxin levels and the development of endometritis in postpartum cows with dystocia

443

or retained placenta. Theriogenology. 2000;54:1019-32.

444

[18] Roche JR, Dillon PG, Stockdale CR, Baumgard LH, VanBaale MJ. Relationships among international

445

body condition scoring systems. J Dairy Sci. 2004;87:3076-9.

446

[19] Barlund CS, Carruthers TD, Waldner CL, Palmer CW. A comparison of diagnostic techniques for

447

postpartum endometritis in dairy cattle. Theriogenology. 2008;69:714-23.

448

[20] McDougall S. Effects of treatment of anestrous dairy cows with gonadotropin-releasing hormone,

449

prostaglandin, and progesterone. J Dairy Sci. 2010;93:1944-59.

450

[21] Quinn PJ, Carter ME, Markey B, Carter GR. Clinical veterinary microbiology. 1st ed. Edinburgh:

451

Harcourt Publishers Limited; 1999.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

429

20

ACCEPTED MANUSCRIPT

[22] Griffin JFT, Hartigan PJ, Nunn WR. Non-specific uterine infection and bovine fertility: I. Infection

453

patterns and endometritis during the first seven weeks post-partum. Theriogenology. 1974;1:91-106.

454

[23] Lin LI-K. A concordance correlation coefficient to evaluate reproducibility. Biometrics. 1989;45:255-

455

68.

456

[24] Picard B, Garcia JS, Gouriou S, Duriez P, Brahimi N, Bingen E, et al. The link between phylogeny and

457

virulence in Escherichia coli extraintestinal infection. Infect Immun. 1999;67:546-53.

458

[25] McNutt LA, Wu CT, Xue XN, Hafner JP. Estimating the relative risk in cohort studies and clinical trials

459

of common outcomes. Am J Epidemiol. 2003;157:940-3.

460

[26] Gilbert RO, Shin ST, Guard CL, Erb HN, Frajblat M. Prevalence of endometritis and its effects on

461

reproductive performance of dairy cows. Theriogenology. 2005;64:1879-88.

462

[27] Kasimanickam R, Duffield TF, Foster RA, Gartley CJ, Leslie KE, Walton JS, et al. Endometrial cytology

463

and ultrasonography for the detection of subclinical endometritis in postpartum dairy cows.

464

Theriogenology. 2004;62:9-23.

465

[28] Hammon DS, Evjen IM, Dhiman TR, Goff JP, Walters JL. Neutrophil function and energy status in

466

Holstein cows with uterine health disorders. Vet Immunol Immunopathol. 2006;113:21-9.

467

[29] Sens A, Heuwieser W. Presence of Escherichia coli, Trueperella pyogenes, -hemolytic streptococci,

468

and coagulase-negative staphylococci and prevalence of subclinical endometritis. J Dairy Sci.

469

2013;96:6347-54.

470

[30] Preziuso S, Cuteri V. A multiplex polymerase chain reaction assay for direct detection and

471

differentiation of -hemolytic streptococci in clinical samples from horses. J Equine Vet Sci. 2012;32:292-

472

6.

473

[31] Madoz LV, Giuliodori MJ, Migliorisi AL, Jaureguiberry M, de la Sota RL. Endometrial cytology, biopsy,

474

and bacteriology for the diagnosis of subclinical endometritis in grazing dairy cows. J Dairy Sci.

475

2014;97:195-201.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

452

21

ACCEPTED MANUSCRIPT

[32] Sheldon IM, Cronin J, Goetze L, Donofrio G, Schuberth H-J. Defining postpartum uterine disease and

477

the mechanisms of infection and immunity in the female reproductive tract in cattle. Biol Reprod.

478

2009;81:1025-32.

479

[33] De Boer MW, Zheng T, Buddle BM, McDougall S. Detection of bovine herpesvirus type 4 antibodies

480

and bovine lymphotropic herpesvirus in New Zealand dairy cows. N Z Vet J. 2014: in press.

481

[34] Deguillaume L, Geffre A, Desquilbet L, Dizien A, Thoumire S, Vorniere C, et al. Effect of endocervical

482

inflammation on days to conception in dairy cows. J Dairy Sci. 2012;95:1776-83.

483

[35] Herath S, Lilly ST, Fischer DP, Williams EJ, Dobson H, Bryant CE, et al. Bacterial lipopolysaccharide

484

induces an endocrine switch from prostaglandin F2 to prostaglandin E2 in bovine endometrium.

485

Endocrinology. 2009;150:1912-20.

486

[36] Xu XJ, Reichner JS, Mastrofrancesco B, Henry WL, Jr., Albina JE. Prostaglandin E2 suppresses

487

lipopolysaccharide-stimulated IFN-beta production. J Immunol. 2008;180:2125-31.

488

[37] Machado VS, Bicalho MLS, Meira EBS, Rossi R, Ribeiro BL, Lima S, et al. Subcutaneous immunization

489

with inactivated bacterial components and purified protein of Escherichia coli, Fusobacterium

490

necrophorum and Trueperella pyogenes prevents puerperal metritis in Holstein dairy cows. PLoS ONE.

491

2014;9:e91734.

492

[38] Gilbert RO, Santos NR, Galvo KN, Brittin SB, Roman HB. The relationship between postpartum

493

uterine bacterial infection (BI) and subclinical endometritis (SE). J Dairy Sci. 2007;90:469-70.

494

[39] Meier S, Priest N, Burke C, Kay J, McDougall S, Mitchell M, et al. Treatment with a nonsteroidal

495

antiinflammatory drug after calving did not improve milk production, health, or reproduction

496

parameters in pasture-grazed dairy cows. J Dairy Sci. 2014;97:2932-43.

497

[40] Clermont O, Bonacorsi S, Bingen E. Rapid and simple determination of the Escherichia coli

498

phylogenetic group. Appl Environ Microbiol. 2000;66:4555-8.

AC
C

EP

TE
D

M
AN
U

SC

RI
PT

476

22

ACCEPTED MANUSCRIPT

[41] Hijazin M, Ulbegi-Mohyla H, Alber J, Lammler C, Hassan AA, Abdulmawjood A, et al. Molecular

500

identification and further characterization of Arcanobacterium pyogenes isolated from bovine mastitis

501

and from various other origins. J Dairy Sci. 2011;94:1813-9.

RI
PT

499

502

AC
C

EP

TE
D

M
AN
U

SC

503

23

ACCEPTED MANUSCRIPT

504
505
506

Table 1. Descriptive statistics of number of cows (%) used for intrauterine sample collection at Day 0
(median 2 d in milk), Day 21 and Day 42 in six commercial spring-calving dairy herds in Waikato, New
Zealand.
Herd
2

515
50

690
50

830
50

3354
272

48

49

45

253

3
5.5
12

3
5
11

3
5
10

3
5
13

16 (33)
6 (13)
26 (54)

0 (0)
45 (92)
4 (8)

40 (89)
0 (0)
5 (11)

117 (46)
56 (22)
80 (32)

1
2
5

1
2
7

1
2
4

1
2
7

7
12
16

3
6
13

3
10
14

3
10
22

9 (19)
10 (21)
3 (6)
24 (50)
88
32
56
6 (13)
40 (95)
28 (58)
27 (59)
38 (79)
46 (96)

4 (8)
12 (25)
3 (6)
19 (39)
98
35
63
14 (29)
33 (94)
30 (61)
27 (57)
39 (80)
47 (96)

3 (7)
4 (9)
4 (9)
27 (60)
92
46
46
3 (7)
36 (86)
23 (51)
23 (59)
33 (73)
41 (91)

24 (10)
37 (15)
18 (7)
137 (54)
NA
NA
NA
54 (21)
177 (89)
150 (59)
141 (61)
198 (78)
239 (95)

AC
C

EP

TE
D

M
AN
U

Herd size
560
329
430
Cows enrolled
50
25
47
Cows included in
47
23
41
analyses
Cow age (years)
Min
3
3
4
Median
6
5
5
Max
13
13
11
Breed
Friesian
31 (66)
20 (87)
10 (24)
Jersey
2 (4)
0 (0)
3 (7)
Crossbreed
14 (30)
3 (13)
28 (68)
b
DIM at Day 0
Min
1
1
1
Median
3
2
2
Max
5
3
5
c
PSC to calving
Min
4
6
8
Median
6
7
14
Max
14
8
22
Diagnosed with
PVDd
Day 0
7 (15)
1 (2)
Day 21
4 (9)
1 (4)
6 (15)
Day 42
5 (11)
1 (4)
2 (5)
CL Day 42
22 (47)
14 (61)
31 (76)
Breedinge (d)
89
96
80
e
AI (d)
40
55
30
e
Bull (d)
49
41
50
Anestrusf
23 (49)
4 (17)
4 (10)
AI 3 wkg
19 (79)
14 (74)
35 (95)
h
Preg 3 wk
30 (64)
14 (61)
25 (61)
Conc to S1 3 wki 29 (69)
14 (78)
21 (54)
h
Preg 6 wk
35 (75)
17 (74)
36 (88)
j
Final preg
43 (92)
22 (96)
40 (98)
a
Number of dairy cows at the start of the calving period

507

Total

RI
PT

1
a

SC

Variable

24

ACCEPTED MANUSCRIPT

DIM = days in milk

509

Days from planned start of the calving period to actual calving of the enrolled cows

510

PVD = Purulent vaginal discharge (flecks of pus or more) detected by MetricheckTM device

511
512

Total length of the breeding program (Breeding; median: 91 d) consisted of a period of AI (median: 37.5
d), followed by natural breeding (Bull; median: 50 d).

513
514

515
516

Cows submitted for AI within the first 3 wk of the start of the breeding program. Cows treated for
anestrus are not included in this variable

517

518

Conception to first service within the first 3 wk of the breeding program

519

Pregnant by the end of breeding program

RI
PT

508

AC
C

EP

TE
D

520

M
AN
U

Pregnant within the first 3 or 6 wk of the breeding program

SC

Cows not detected in estrus by tail paint removal within one month prior to the start of the breeding
program

25

ACCEPTED MANUSCRIPT

Table 2. PCR primers used for the phylogenetic grouping of E. coli (chuA, yjaA, and TSPE4.C2) and the identification of T. pyogenes (plo, sodA,

522

and ISR).

152

SC

211

270
199

Hijazin et al., 2011

[41][42][42]
Ulbegi-Mohyla et al., 2010
[43]

TE
D

122

References
Clermont et al., 2000
[40][41][40]
Clermont et al., 2000
[40][41][40]
Clermont et al., 2000
[40][41][40]
Jost et al., 2002 [41]

EP

524

Size of PCR
product (bp)
279

AC
C

523

Direction
Sequence (5 3)
Forward
GACGAACCAACGGTCAGGAT
Reverse
TGCCGCCAGTACCAAAGACA
yjaA
Forward
TGAAGTGTCAGGAGACGCTG
Reverse
ATGGAGAATGCGTTCCTCAAC
TSPE4.C2
Forward
GAGTAATGTCGGGGCATTCA
Reverse
CGCGCCAACAAAGTATTACG
plo
Forward
GGCCCGAATGTCACCGC
Reverse
AACTCCGCCTCTAGCGC
sodA
Forward
CGAGCTCGCCGACGCTATTGCT
Reverse
GAGCATGAGAATCGGGTAAGTGCCA
ISRa
Forward
GTTTTGCTTGTGATCGTGGTGGTTATGA
Reverse
AAGCAGGCCCACGCGCAGG
a
ISR = 16S-23S rDNA intergenic spacer region

M
AN
U

Primer
chuA

RI
PT

521

26

ACCEPTED MANUSCRIPT

Table 3. Number of cows (n) and proportion (% of cows enrolled in that herd) with E. coli, T. pyogenes or any bacterial growth irrespective of

526

species (Any bacteria) at Days 0 (median 2 d in milk) and 21, by study herd. Between herd differences calculated from multivariable models with

527

Herd 1 as the referent herd: #P 0.10, *P 0.05, **P 0.01 and ***P 0.001.

AC
C

EP

TE
D

528

Any bacteria (n; %)

Day 0
Day 21
45 (96) 42 (89)
20 (87) 17 (74)
40 (98) 34 (83)
43 (90)* 35 (73)#
44 (90)# 41 (84)
43 (96) 33 (73)#
235 (93) 202 (79)

SC

Day 21
13 (28)
4 (17)
7 (17)
7 (15)
18 (37)
10 (22)
59 (24)

T. pyogenes (n; %)
Day 0
Day 21
2 (4)
8 (17)
0 (0)
1 (4)#
0 (0)
3 (7)
1 (2)
3 (6)
0 (0)
4 (8)
0 (0)
4 (9)
3 (1)
23 (9)

M
AN
U

E. coli (n; %)
Herd Cows (n) Day 0
1
47
32 (68)
2
23
5 (22)***
3
41
10 (24)***
4
48
20 (42)*
5
49
15 (31)***
6
45
23 (51)
Total
253
105 (42)

RI
PT

525

27

ACCEPTED MANUSCRIPT

Table 4. Number of cows (n) and proportion (%) among cows with E. coli isolated from the uterus stratified by phylogenetic groups, at Days 0
(median 2 d in milk) and 21 postpartum, by study herd.

32
5
10
20
15
23
105

A
6 (19)
1 (20)
5 (50)
4 (20)
8 (53)
5 (22)
29 (28)

B1
14 (44)
3 (60)
4 (40)
14 (70)
6 (40)
16 (70)
57 (55)

B2
2 (6)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
2 (2)

D
10 (31)
1 (20)
1 (10)
2 (10)
1 (7)
2 (9)
17 (15)

13
4
7
7
18
10
59

A
4 (31)
0 (0)
5 (71)
1 (14)
6 (33)
4 (40)
20 (33)

B1
7 (54)
4 (100)
2 (29)
4 (57)
8 (44)
5 (50)
30 (51)

B2
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)

D
2 (15)
0 (0)
0 (0)
2 (29)
4 (22)
1 (10)
9 (15)

AC
C

EP

TE
D

531

Cows with E.
coli (n)

SC

Herd
1
2
3
4
5
6
Total

Day 21
Phylogenetic group (n; %)

RI
PT

Day 0
Phylogenetic group (n; %)

Cows with E.
coli (n)

M
AN
U

529
530

28

ACCEPTED MANUSCRIPT

532
Fig. 1. Study design to assess associations between intrauterine bacterial isolation (Bacteriology),

534

reproductive tract inflammation as diagnosed by Metricheck and cytobrush (Cytology), and reproductive

535

performance in pasture-based dairy cows (n = 272) from six herds bred from the start of the seasonal

536

breeding program (planned start of mating; PSM)

AC
C

EP

TE
D

M
AN
U

SC

537

RI
PT

533

29

ACCEPTED MANUSCRIPT

538
Fig. 2. Diagram of associations between-sample days (i.e. Days 0, 21 and 42) found between E. coli

540

(including phylogenetic groups; group) and T. pyogenes (T. pyo) isolated from the uterus, culture

541

positive irrespective of bacterial species (bacteria), purulent vaginal discharge (PVD), percentage

542

polymorphonuclear cells (PMN%), anestrus treatment (anoestrus) and reproductive outcomes of 253

543

dairy cows from six herds. Each association is presented by relative risk (95% CI), except when specified

544

as -coefficient (-coef; 95% CI) from linear regression or hazard ratio (HR; 95% CI) from Coxs

545

proportional hazard regression. P-values are giving by the weight of the arrows (dashed: P 0.10, solid

546

thin: P 0.05, and solid thick: P 0.01).

M
AN
U

SC

RI
PT

539

AC
C

EP

TE
D

547

30

ACCEPTED MANUSCRIPT

548
Fig. 3. Frequency distribution of the percentage of intrauterine polymorphonuclear cells (PMN%)

550

diagnosed at (a) Day 0 and (b) Day 21 from 253 cows from six New Zealand dairy herds.

RI
PT

549

AC
C

EP

TE
D

M
AN
U

SC

551

31

ACCEPTED MANUSCRIPT

552
Fig. 4. Survival curves for time to pregnancy up to the end of the seasonal breeding program in 253 dairy

554

cows from six herds and categorized by intrauterine presence of E. coli (n = 59) at Day 21 (median: 23

555

DIM). Median days from the start of the seasonal breeding program to pregnancy (95% CI; P-value) were

556

14 (12-18) and 17 (9-24; 0.08) for cows without and with E. coli isolation, respectively.

RI
PT

553

AC
C

EP

TE
D

M
AN
U

SC

557

32

ACCEPTED MANUSCRIPT

558
Fig. 5. Survival curves for time to pregnancy up to the end of the seasonal breeding program in 253 dairy

560

cows from six herds and categorized by affected by polymorphonuclear cells (PMN; 25%; n = 157) at

561

Day 0 (median: 2 d in milk). Median days from the start of the seasonal breeding program to pregnancy

562

(95% CI; P-value) were 16 (12-24) and 14 (11-18; 0.05) for cows with <25% and those with 25% PMN,

563

respectively.

SC

RI
PT

559

AC
C

EP

TE
D

M
AN
U

564

33

Pregnancy
test

Anoestrus
+ treatment

Day 42
( 2 d)

PSM
-10 d

PSM

TE
D

Day 21
( 0 d)

Pregnancy
test

Bull

M
AN
U

AI

SC

Metricheck
+ treatment
BCS
CL +/-

EP

Day 0

Metricheck
Cytology
Bacteriology
BCS
Uterine tone

AC
C

Metricheck
Cytology
Bacteriology
BCS
Uterine tone
Rectal
temperature

RI
PT

ACCEPTED MANUSCRIPT

PSM +
11-14 wk

Bull out +
5 wk

ACCEPTED MANUSCRIPT

1.16 (1.02-1.31)

Day 42
0.78 (0.63-0.97)

Bacteria

CL +ve

0.60 (0.40-0.91)
group A
1.31 (1.03-1.67)

E. coli

0.20 (0.03-1.36)

T. pyo

T. pyo

-coef:
0.10 (0.03-0.17)

PVD

1.02 (1.00-1.04)

PMN%

AC
C

PMN%

TE
D

1.02 (1.01-1.03)

PVD

EP

PVD

Pregnancy
3 wk
Anoestrus

2.99 (1.16-7.75)
1.57 (1.01-2.44)

1.89
(1.04-3.43)

0.61 (0.38-0.97)

AI 3 wk

HR: 0.75 (0.54-1.30)

M
AN
U

E. coli

1.97 (1.26-3.10)

Reproduction

RI
PT

Bacteria

Day 21

SC

Day 0

Conception to
1st AI 3 wk
Pregnancy
6 wk

0.99 (0.99-1.00)

HR (25%): 1.32 (1.00-1.74)

Time to
pregnancy

10

Frequency
20
30

40

AC
C

50

EP

TE
D

M
AN
U

SC

RI
PT

ACCEPTED MANUSCRIPT

20

40
60
PMN% at Day 0

80

100

10

Frequency
20
30

40

AC
C

50

EP

TE
D

M
AN
U

SC

RI
PT

ACCEPTED MANUSCRIPT

20

40
60
PMN% at Day 21

80

100

E. coli ve at Day 21
E. coli +ve at Day 21

0.00

AC
C

Cumulative proportion of cows not pregnant

0.25
0.50
0.75
1.00

EP

TE
D

M
AN
U

SC

RI
PT

ACCEPTED MANUSCRIPT

20

40
60
Time from PSM to pregnancy (days)

80

100

< 25% PMN at Day 0

25% PMN at Day 0

0.00

AC
C

Cumulative proportion of cows not pregnant

0.25
0.50
0.75
1.00

EP

TE
D

M
AN
U

SC

RI
PT

ACCEPTED MANUSCRIPT

20

40
60
Time from PSM to pregnancy (days)

80

100

ACCEPTED MANUSCRIPT

Highlights
Presence of intrauterine E. coli at 2 DIM did not predispose T. pyogenes in cows

E. coli at 2 DIM increased the risk of E. coli or any bacteria 21 days later

E. coli at 2 DIM and T. pyogenes at 23 DIM increased risk of PVD 21 days later

Any bacteria at 23 DIM decreased the risk of pregnancy within 3 weeks from

RI
PT

EP

TE
D

M
AN
U

Cows with 25% PMN detected by cytobrush at 2 DIM had a shorter time to pregnancy

AC
C

SC

breeding