Coliform mastitis in sows: Analysis of potential ... · pig production, and has been reported...

Aus dem Institut für Tierzucht und Tierhaltung der Agrar- und Ernährungswissenschaftlichen Fakultät

der Christian-Albrechts-Universität zu Kiel

Coliform mastitis in sows:

Analysis of potential influencing factors and bacterial pathogens with special emphasis on

Escherichia coli

Dissertation

zur Erlangung des Doktorgrades der Agrar- und Ernährungswissenschaftlichen Fakultät

der Christian-Albrechts-Universität zu Kiel

vorgelegt von

Master of Sciene

Imke Gerjets

aus Wittmund, Niedersachsen

Dekanin: Prof. Dr. K. Schwarz Erste Berichterstatterin: Prof. Dr. med. vet. N. Kemper

Zweiter Berichterstatter: Prof. Dr. J. Krieter

Tag der mündlichen Prüfung: 10.02.2011

Die Dissertation wurde mit dankenswerter finanzieller Unterstützung des Bundesministeriums für Bildung und Forschung (BMBF) im Rahmen der FUGATO-plus-Nachwuchsgruppe ‚geMMA: structural and functional analysis of the genetic

variation of the MMA-syndrome’ angefertigt.

Table of Contents

General Introduction ................................................................................................................ 1

Chapter 1 .................................................................................................................................. 3

Coliform mastitis in sows: A review .................................................................................. 3

Chapter 2 ................................................................................................................................ 32

Comparison of virulence gene profiles of Escherichia coli isolates from sows with

Coliform mastitis and healthy sows ................................................................................ 32

Chapter 3 ................................................................................................................................ 50

Assessing individual sow risk factors for coliform mastitis in sows:

A case-control study .......................................................................................................... 50

Chapter 4 ................................................................................................................................ 61

Application of decision-tree technique to assess herd specific risk factors for

coliform mastitis in sows ................................................................................................... 61

General Discussion ............................................................................................................... 75

General Summary ................................................................................................................. 82

Zusammenfassung ................................................................................................................ 84

1

General Introduction

The survival and growth of piglets in their first days of life is strongly dependent on

adequate colostrum and milk production by the sow. Coliform mastitis (CM), a

disease in sows occurring after farrowing, is not only characterised by fever and an

inflammation of the mammary glands, but also, as a consequence, by greatly

reduced milk production within 12 to 48 hours post-partum. The syndrome affects

therefore the productivity of the sows as well as the growth and the preweaning

mortality of the piglets. It is a serious problem for the economy and animal welfare in

pig production, and has been reported worldwide. Coliform mastitis is a multifactorial

disease, and most research to date has focused on the husbandry-influenced

occurrence, although a single pathway is unlikely to exist. In pathogenesis, there are

hints of a predominant influence of Escherichia (E.) coli.

The aim of this thesis was the functional phenotyping of CM-affected sows involving

advanced bacteriological techniques. All bacteria isolated from milk samples of

diseased and healthy sows were identified at an extensive level and compared.

Special emphasis was given to E. coli isolates.

A further objective was the analysis of sow- and birth-related factors contributing to

the occurrence of CM, which were assessed under production conditions. Different

statistical approaches were applied.

The first chapter provides an insight into the disease complex according to present

knowledge. Most studies on the topic were carried out between 1970 and 1990.

Changes in pig production over the last few decades and the still existing economic

losses demand a closer look at CM again.

Chapter two deals with the analysis of milk samples of sows with and without CM for

the presence of E. coli. All identified E. coli isolates were subsequently investigated

for particular virulence genes, including genes for adhesion factors, toxins, iron

acquisition factors, lipopolysaccharides, polysaccharide capsules and invasion

factors. New findings on the pathogenesis of the disease and the occurrence of

different virulence factors in E. coli isolates associated with coliform mastitis in sows

were to be attained.

2

The focus of chapter three is on identifying potential risk factors, in particular

individual sow characteristics, for CM by a case-control study. In this epidemiological

clinical study, diseased sows were matched with healthy ones of the same herd by

conditional logistic regression. In addition, a second case-control study was

conducted to investigate the risk of repeated clinical mastitides in the following

lactations for sows that had already suffered mastitis before.

In chapter four, the application of the decision-tree technique to potential risk factors

for CM, analysed in chapter three, is investigated. The aim of this data mining

method was to make sow herd datasets accessible and comparable by generating

graphical trees and by visualizing possible decision rules. The ability of the decision-

tree technique to distinguish between sows with CM from healthy ones and to predict

the outcome of the disease was analysed.

3

Chapter 1

Coliform mastitis in sows: A review

Imke Gerjets, M. sc. agr.1

Nicole Kemper, Prof. Dr. med. vet.2

1Institute of Animal Breeding and Husbandry, Christian-Albrechts-University Kiel, D-

24098 Kiel, Germany

2Institute of Agricultural and Nutritional Sciences, Martin-Luther-University Halle-

Wittenberg, D-06120 Halle, Germany

Journal of Swine Health and Production (2009) 17 (2): 97-105

4

Abstract

Coliform mastitis (CM) represents an economically very important disease complex in

sows that also affects the health, welfare, and performance of the piglets. Most

research has concentrated on the husbandry-influenced occurrence of CM. In

pathogenesis, there are many hints of an outstanding influence of Escherichia coli

and its endotoxins, although different species among the Enterobacteriaceae have

been isolated from affected animals. Most studies on this topic were conducted

between 1970 and 1990. But with particular respect to the economic damages and

the lack of recent literature, it is time for research to have a closer look at this disease

again. The use of body temperature as a single indicator for CM diagnosis and

treatment must be regarded critically. To minimise use of antibiotics and to achieve a

proper diagnosis, a combination of appropriate criteria should be applied. Additional

approaches, for instance, genetic resistance, are promising tools for future

prevention.

KEYWORDS

Swine, mastitis, dysgalactia, sows, endotoxins

5

Introduction

Postparturient disorders represent an economically important disease complex in

sows world-wide (Bertschinger, 1999), incurring losses due to reduced productivity

and high mortality rates. These disorders are commonly categorized under the terms

mastitis-metritis-agalactia (MMA) complex (Martin et al., 1967), postpartum

dysgalactia syndrome (PPDS or PDS) (Klopfenstein et al., 2006), and periparturient

hypogalactia syndrome (Smith et al., 1992). Miscellaneous other names, such as

agalactia complex (Penny, 1970), lactation failure (Elmore and Martin, 1980),

agalactia toxemica (Ringarp, 1960), or agalactia postpartum syndrome (Hermannson

et al., 1978), reflect the numerous aetiologies involved in the pathophysiology of this

disease that varies in its clinical presentation. All these terms summarize the

characteristic syndrome of greatly reduced milk production within 12 to 48 hours

postpartum, leading rapidly to piglet starvation. However, the name MMA complex

is misleading, as metritis is found only occasionally in affected animals (Heinritzi and

Hagn, 1999; Waldmann and Wendt, 2001), and instead of total agalactia, sows

continue to produce milk at a reduced level. Still, MMA is the commonly used term in

European countries, while PPDS or PDS have become widely accepted in English-

speaking areas (Waldmann and Wendt, 2001; Klopfenstein et al., 2006).

Of the variety of conditions related to puerperal disorders in sows, mastitis is one of

the central clinical signs, as shown by several studies (Ross et al., 1981; Wegmann

et al., 1986; Heinritzi and Hagn, 1999). Bacteria most commonly isolated from

affected sows are coliforms, including the genera Escherichia, Citrobacter,

Enterobacter, and Klebsiella (Ross et al., 1981; Wegmann et al., 1986; Awad

Masalmeh et al., 1990; Hirsch et al., 2003; Gerjets et al., 2008). The predominant role

of these organisms in mastitis in sows has been demonstrated by several infection

experiments (Ross et al., 1981; Wegmann and Bertschinger, 1984; Bertschinger et

al., 1990). Hence, to avoid the confusing terminology and to point out the parallels to

coliform mastitis in cows, the term coliform mastitis (CM) was suggested for peripartal

mastitis in sows (Bertschinger, 1999). This review will concentrate on CM as an

essential part of the puerperal disease complex and as a major cause of dysgalactia

in sows. Most investigations into CM were carried out between 1970 and 1990, and

the scarcity of recent studies is reflected in the reference list of this review.

6

As shown in Sweden, udder problems are the reason for culling up to 13% of sows

(Ringmar-Cederberg and Johnson, 1996), but the main adverse economic effect of

CM is a high pre-weaning piglet mortality (Furniss, 1987). The piglets are totally

reliant on the sow for access to colostrum and milk, and growth rate depends both on

milk yield and composition (Grün et al., 1993). By lying on their mammary glands,

affected sows refuse piglets access to the teats. As a result of dysgalactia in

combination with pain in the mammary gland, the sow fails to meet the needs of the

piglets. Mortality and growth retardation in piglets are the result (Ringarp, 1960;

Penny, 1970). The first 3 days after birth are the most critical period for survival of

piglets. As glycogen stores are very low in new-born piglets and glyconeogenesis is

insufficient, hypoglycemia may be induced in piglets with insufficient milk intake by

the rapid decrease in glycogen (Sujatha et al., 2003). Inadequate colostrum intake

results in deaths primarily due to starvation and hypothermia, but also because of

inadequate transfer of maternal immunoglobulins to the piglet. Due to its energy and

immunoglobulin content, a sufficient intake of colostrum is essential for healthy

development of piglets. Inadequate colostrum intake is often followed by severe

health problems, for instance, diarrhea, poor growth, and inanition (Rooke and Bland,

2002). Thus, CM creates animal welfare issues both for the sow and her piglets.

Even though infection is not transmitted through animal-animal contact, CM may

become nearly epidemic in affected herds, with up to 80% of sows affected

(Waldmann, 2000). In other herds, it may be limited to a few animals and may be

only sporadic. The incidence of CM at farm level is reported to vary from 0.5% to

60% (Hirsch et al., 2004) in Scandinavia and from 1.1% and 37.2% (Bäckström et al.,

1984) in Illinois; but average incidence at herd level is approximately 13%

(Hermannson et al., 1978; Jorsal, 1983; Bäckström et al., 1984; Madec and Leon,

1992; Thorup, 2000; Krieter and Presuhn, 2005). Herds managed using totally

different hygienic practices and standards may be affected (Waldmann and Wendt,

2001; Hirsch et al., 2003); CM even occurs on excellently managed farms with

optimized disinfection practices (Gerjets et al., 2008; Papadopoulos et al., 2008).

7

Pathological findings

In recent years, several attempts have been made to classify the wide variety of

clinical syndromes affecting the sow’s mammary gland diagnosed in the peripartal

period, but no classification has become widely accepted. For example,

classifications have been based on the number of affected glands, including

uniglandular or multiglandular mastitis, or, with regard to duration and state of

inflammation, mastitis has been subdivided into acute and chronic mastitis

(Waldmann and Wendt, 2001). Systemic signs of disease, such as fever and

anorexia, are widespread, often associated with constipation and depression (Scuka

et al., 2006a). The infected glands show typical signs of inflammation, such as severe

oedema and skin congestion. There may be acute induration of the mammary region,

although oedema without signs of acute mastitis can be found, especially in

primiparous sows (Martin et al., 1967; Nachreiner et al., 1971). Caudal glands are

reported to be more affected than cranial ones (Bostedt et al., 1998b), but in contrast,

a more recent study detected no differences with regard to anatomic location (Gerjets

et al., 2008). Other pathological findings may include fever, constipation, vulvovaginal

discharge, skin discoloration, and anorexia. Haematological findings comprise

leucopenia or leucocytosis, a decrease in packed cell volume and haemoglobin

concentration, and an increase in serum phosphorus concentration, while

concentrations of serum calcium, magnesium, and glucose may decrease (Baer and

Bilkei, 2005).

A histological study by Swarbrick (1968) revealed an accumulation of secretion in

mammary glands of affected sows. These findings, and the fact that early initiation of

lactation (up to 24 hours before parturition) might result in engorgement of the

mammary gland, suggest that early lactation is a predisposing cause of CM (Martin et

al., 1978; Gooneratne et al., 1982). Initiation of lactation is induced by a decline in

plasma progesterone level (Kuhn, 1969), which may appear earlier in sows with CM

(Gooneratne et al., 1982). In contrast, a delayed decline in plasma progesterone

level was reported by Liptrap (1980) as a causative factor for development of clinical

CM.

In piglets, reduction of milk intake causes various clinical signs. The greater tendency

of the sow to lie in lateral recumbency, combined with the weakness of malnourished

piglets, results in an increased incidence of crushing (Hellbrugge et al., 2008). Total

8

piglet mortality up to the age of 1 week in the litters of CM-affected sows varies from

5.0% (Hühn and Rehbock, 2008) to 38.6% (Bäckström et al., 1984). In a study with

46 sows, the mammary secretion of sows that subsequently developed CM within 12

to 24 hours after farrowing contained significantly higher concentrations of lactose

and significantly lower concentrations of protein and Na+ compared to milk from

unaffected sows, while the concentration of fat and K+ was similar (Gooneratne et al.,

1982). From these results, the authors of this study suggested an analysis of

colostrum to identify sows predisposed to CM, to indicate affected glands, and to

monitor recovery, but as this was not put into practice, there is a lack of further

evidence for this theory. Coliform mastitis is often followed by temporary or

permanent infertility (Bilkei et al., 1994a) caused by direct bacterial and inflammatory

effects on the genital tract that prevent conception. A direct effect on the onset of the

estrus cycle may not be important for development of later infertility (ten Napel et al.,

1995).

Diagnosis

Diagnosis of CM in commercial herds is based mainly on clinical signs. Hypogalactia

within the first 3 days post-partum suggests CM (Bertschinger, 1999). Piglets make

vigorous nursing efforts. Both the decrease in nursing intervals and the increase in

piglets’ activity derive from absent or reduced milk ejection (Bertschinger, 1999). The

piglet’s strenuous nursing efforts may cause traumatized teats. After exhaustion of

their energy reserves, piglets often retreat to the warmest parts of the farrowing crate

and decrease their attempts to nurse (Klopfenstein et al., 2006). In sows, mammary

glands may appear normal or pathologically altered, varying from swollen, firm, and

warm to the touch. In addition, the skin colour can be changed.

After studying the relationship between elevated temperature and CM, it was

proposed to use post-farrowing rectal temperature to determine whether CM was

likely to become a serious problem (Larsen and Thorup, 2006). A study by

Hermansson et al. (1978), comparing 71 sows affected with mastitis to 71 healthy

sows, showed a significantly higher body temperature for the affected ones. The first

trial to evaluate sow rectal temperature as a predictor of CM and to determine the

specific time when the sow’s temperature should be taken was conducted by Furniss

(1987). This study suggested that a rectal temperature of 39.4°C occurring 12 to 18

9

hours after farrowing is an appropriate threshold at which to give preventive

treatment. Today, the most common practice used to detect an animal’s risk of CM is

to measure the rectal temperature post-partum. Besides abnormal temperature,

criteria for the diagnosis of CM must include the combination of clinical mammary

gland changes, diminished milk production, and reduced appetite (Mirko and Bilkei,

2004). The range of critical temperature values varies between 39.3°C and 40.5°C

(Waldmann and Wendt, 2001), but physiological hyperthermia is often observed in

postparturient sows, leading to misinterpretations (Klopfenstein et al., 2006; Gerjets

et al., 2008).

Body temperature is a non-specific parameter indicating alterations of the

physiological state of warm-blooded animals. Plasma concentrations of acute phase

proteins such as α1-acid-glycoprotein (AGP) and haptoglobin (Hp), which are part of

the immune system, increase in stressful situations and can be used as indicators of

acute CM (Mirko and Bilkei, 2004). Plasma concentrations of cortisol and 15-

ketodihydroxy-PGF2α have also been suggested as inflammation indicators (Garcia

et al., 1998). All of these parameters can vary substantially at the time of parturition

(Magnusson and Fossum, 1992; Österlundh et al., 2002), and as collecting blood

samples is much more laborious than measuring body temperature, use of such

nonspecific indicators to diagnose CM is not feasible under field conditions. Another

attempt to diagnose puerperal diseases in sows at a very early stage was made by

Petersen (1983), who suggested the combination of several urine parameters to

diagnose bacteriuria. In a further study, it has been shown that analysis of urinary

concentrations of minerals, especially potassium, in urine samples collected from

sows in the morning and afternoon during mid-lactation provide an acceptable

estimation of milk production (Papadopoulos et al., 2007).

Baer and Bilkei (2005) investigated the use of ultrasonography for differentiating

sows having suffered recurrent CM from healthy animals. It was shown that with a

linear array technique and a frequency of 8.5 MHz, affected mammary glands provide

hyperechogenic images. Furthermore, this study supports the theory that abdominal

glands are more prone to pathological changes than the pectoral glands. The use of

ultrasonography as a precautionary measure has not been integrated into herd

management due to impractical handling and additional costs.

10

Rapid mastitis tests as applied to cows are not commercially available for sows.

Diagnosis via cell count is not common and data on thresholds are rare. For instance,

a threshold of 5 × 106 cells per mL was proposed by Bertschinger and Bühlmann

(1990), while Persson et al. (1996a) suggested 10 × 106 cells per mL. All parameters

used to detect CM are summarized in Table 1.

Table 1: Parameters altered in CM-affected sows

Parameter CM sows Literature cited

Body temperature

> 39.3°C Hoy (2003)

> 39.5°C Furniss (1987)

> 40.0°C Kiss and Bilkei (2005)

Milk production Hypogalactia, dysgalactia, agalactia Kiss and Bilkei (2005)

Appetite Diminished; moderate or total anorexia Kiss and Bilkei (2005)

Cell count > 107/mL Waldmann and Wendt (2001)

Milk pH > 6.7 Waldmann and Wendt (2001)

Urine parameters Bacteriuria and proteinuria Petersen (1983)

Interleukines Increased IL-1ß, IL-6, IL8, and TNFα Zhu et al. (2007b)

Factors influencing clinical CM

The aetiology of CM seems to be inconsistent and challenging. Indeed, the

occurrence of the disease is multifactorial. The anatomy of the sows’ mammary

glands is different from that of cows. Two complete gland systems end in two teat

orifices per teat, without muscular sphincters (Klopfenstein et al., 2006). The gland

cisterns are not well-defined. During the last part of each gestation, mammogenesis

recurs, which implies that new glandular tissue is produced. This results in a great

ability of the sow to restore mammary health from one lactation to the next, although

chronic lesions of the teat canal are usually irreversible (Hartmann et al., 1997;

Hurley, 2001).

Coliform bacteria are ubiquitous, and therefore, influence the factors that determine

the development of infection in the single animal. Factors contributing to clinically

apparent CM include the strongly related main issues of nutrition, housing

microclimate, management in general, and aspects of hygiene in particular. The

11

factors identified thus far for an increase in CM prevalence are summarized in Table

2.

Table 2: Non-infectious factors increasing the occurrence of CM

Factor Literature cited

At individual level

Sows of higher parity (> 4) Baer and Bilkei (2005)

Young sows of lower parity (1.2) Bostedt et al. (1998b), Hoy (2002), Krieter and

Presuhn (2005)

Long gestation > 116 days Awad Masalmeh (1990)

Long duration of birth (> 3 hours) Bostedt et al. (1998b)

After obstetric intervention Bostedt et al. (1998b)

Large litter size (> 11) Bostedt et al. (1998b)

Urinary tract infections Berner (1971), Petersen (1983)

Obstipation Bostedt et al. (1998b)

Genetic disposition Awad Masalmeh (1990)

At herd level

Increasing herd size Bäckström et al. (1984)

Smaller herd size Ringarp (1960)

Change of housing Waldmann and Wendt (2001)

In new herds of gilts Waldmann and Wendt (2001)

Seasonal influences Awad Masalmeh (1990)

Lack of crude fiber in the ration Plonait and Bickhart (1997)

Rapid changes in nutrition Plonait and Bickhart (1997)

Single housing, lack of exercise Hoy (2002), Ringarp (1960)

Information about the influence of parity number on occurrence of CM is contradictory

(Berner, 1971; Bostedt et al., 1998a; Baer and Bilkei, 2005). The normal length of

gestation in sows varies between 113 and 117 days, and CM often occurs in sows

with a gestation of > 116 days (Awad Masalmeh et al., 1990). All factors contributing

to prolonged duration of the birth process increase the prevalence of CM (Berner,

1971; Bostedt et al., 1998b; Papadopoulos et al., 2007), as does the concurrent

occurrence of urinary tract infections (Bilkei et al., 1994a). Nutrition clearly impacts

the fertility of sows at various points in the life of the sow. Several factors, such as

imbalanced diet, lack of fibre, excessive feeding, or mycotoxins (i.e., in mouldy feed),

must be taken into account (Heinritzi et al., 2006; Scuka et al., 2006a).

12

Obstipation due to diet and inadequate water intake creates further risk of CM,

probably by increasing the endogenous transfer of bacteria and endotoxins to the

mammary gland (Waldmann and Wendt, 2001; Krüger et al., 2002). The influence of

nutrition on the hypothalamo-hypophysical gonadal axis was evaluated in a review by

Cosgrove and Foxcroft (1996), who emphasized the importance of appropriate

nutritional management to support the endocrine system and its influence on

lactogenesis. Seasonal influences are largely eliminated by the circumstances of

modern production (Bilkei et al., 1994b). However, high ambient temperatures may

cause stress responses in sows, with a negative effect on reproductive performance.

During lactation, high ambient temperature (> 27°C) may reduce voluntary food

intake and enhance lactational weight loss (ten Napel et al., 1995; Prunier et al.,

1997). This results in a contradiction for swine management in intensive piggeries:

the ideal temperature for the sow to exploit her full lactation potential (< 24°C) is not

the ideal temperature for her piglets (> 30°C) (Hartmann et al., 1997). The

significance of these influences has been considered in management practices by

providing heat lamps and other heating devices in the creep area. Late introduction

into the farrowing pen, i.e., after the 110th day of gestation, is associated with an

increase in CM prevalence (Scuka et al., 2006b). Furthermore, a tendency towards a

lower prevalence of CM with increasing herd size was observed (Lingaas and

Ronningen, 1991). In contrast to this, Bäckström et al. (1984) found a higher

prevalence of CM with increasing herd size.

Bacteria and endotoxins causing CM

The causative agents of CM and their role in pathogenesis have been discussed

controversially, as many different bacterial species have been isolated from the milk

of clinically diseased animals (Awad Masalmeh et al., 1990; Kobera, 2000), including

mainly coliform bacteria (Escherichia coli and other lactose-splitting bacteria), but

also Streptococci, Staphylococci, Pseudomonas species, and Corynebacterium

species. One problem regarding the presence of different bacterial species in the milk

of affected animals is the use of inadequate methods for identification.

There are three main theories concerning the routes of infection for CM: endogenous,

including the gut and the uterus, and exogenous via the mammary gland. The

infectious dose for colonization of the mammary gland is extremely low at < 100

13

organisms (Österlundh et al., 2002; Papadopoulos et al., 2007). Causative bacteria

are located free in the milk or in phagocytic cells in the ductular and alveolar lumina

and are often isolated from regional lymph nodes (Armstrong et al., 1968;

Bertschinger et al., 1977a; Ross et al., 1981). In a study comparing the bacterial flora

of the uterus, the cecum, the ileum, and the mammary gland in order to identify a

likely source of endotoxin absorption, the prevalence of only gram-negative bacteria

in the mammary glands and in the ileum of CM-affected sows was remarkable

(Morkoc et al., 1983). The lack of gram-negative bacterial culture growth in uterine

samples supports the theory that uterine involvement in CM is of minor importance,

as has been suggested (Armstrong et al., 1968; Martin, 1970; Nachreiner and

Ginther, 1974). The hypothesis of a galactogenous route of infection via the teat duct

is supported by experiments carried out by Bertschinger et al. (1990) and

Bertschinger et al. (1977b), who found a lower prevalence of CM when the mammary

gland was protected against faecal contamination. Due to repeated sampling, the

time of infection could be determined in this experimental setting. More than 50% of

mammary glands were infected before parturition, but no new infections appeared

before the 108th day of gestation (Bertschinger et al., 1990). New infections were

limited to the first 2 days after farrowing. This limitation was explained by the

established teat preference of the piglets and suckling at regular intervals of three-

quarters of an hour (Bertschinger et al., 1990).

All isolated gram-negative bacteria are common in the sows’ environment, depending

on a combination of circumstances. For instance, the use of wood shavings as

bedding material leads to an increased occurrence of pathogenic Klebsiella

pneumoniae (Hogan and Smith, 1997), that might end in more infections of the

mammary glands of the sows due to a high contamination rate in the material. The

origin of bacteria in the environment may be related to the excretion of urine and

faeces by the sows. In this context, it is notable that infections of the urinary tract are

strongly related to puerperal diseases, even though urinary infections are not

apparent clinically (Mauch and Bilkei, 2004). The most common organism associated

with bacteriuria and vulval discharge was found to be E coli (Waller et al., 2002). The

mammary gland as a source of gram-negative bacteria was first described by Elmore

et al. (1978) and Jones (1979). The predominant role of coliform bacteria in

pathogenesis was clearly shown by Wegmann et al. (1986); both E coli and K

pneumoniae were isolated from 79% of 131 mammary complexes of CM-affected

14

sows. In a study with 663 sows suffering recurrent CM, bacteriological examination of

mammary gland changes revealed the presence of mainly E coli and Klebsiella

species, but also Clostridium species, Actinobaculum suis, Pseudomonas

aeruginosa, Proteus species, gram-positive streptococci (especially Enterococci and

Streptococcus faecalis), staphylococci (Staphylococcus albus, Staphylococcus

epidermis, Staphylococcus aureus), and Erysipelothrix rhusiopathiae (Baer and

Bilkei, 2005).

The prominent role of E coli in mastitis has been emphasized in several studies

(Armstrong et al., 1968; Bertschinger et al., 1977a; Ross et al., 1981; Wegmann et

al., 1986). Bacteriological examinations of milk and udder biopsies and necropsy

material from sows with CM have indicated that E coli is the causative pathogen for

agalactia in the majority of cases (Persson, 1997; Pedersen Mörner et al., 1998).

Typically, peripartum mastitis caused by E coli is acute (Bäckström et al., 1984), but

postparturient mastitis has also been described in sows lacking signs of clinical CM

(Persson, 1997). In sows, experimentally induced E coli or K pneumoniae mastitis

provokes clinical and haematological changes comparable to natural infections

(Bertschinger et al., 1977a; Ross et al., 1983). The extensive interplay between

pathogen and host can cause different clinical syndromes. While some sows develop

clinical signs of CM after inoculation of the mammary glands with E coli, others

remain unaffected (Österlundh et al., 1998). A large study of 39 pairs of full siblings

(Swedish Landrace × Swedish Yorkshire) over six parities demonstrated that less

than half of the mammary glands with CM (diagnosed by milk bacteriology and

cytology) showed clinically detectable mastitis (Persson et al., 1996b).

Nevertheless, the involvement of defined E coli strains and the occurrence of certain

virulence determinants such as shigatoxins remain ambiguous with regard to the

development of clinical appearance (Pedersen Mörner et al., 1998). A wide variety of

E coli serotypes have been substantiated in mastitic sows’ milk in previous studies

(Armstrong et al., 1968; Awad Masalmeh et al., 1990). Bostedt et al. (1998b) found a

high percentage of antibiotic-resistant E coli in cervical swabs from sows with CM:

the isolated strains were 100% sensitive only to gentamicin. Sensitivity to all other

tested antibiotics was < 100%. The findings of Pedersen Mörner et al. (1998) support

the theory of a galactogenous route of infection: serological homogeneity was found

15

in E coli isolates from the same teats at different times during lactation, while

heterogeneity was encountered for different teats in the same sampling. On the basis

of current knowledge, this may be interpreted as mastitis in sows being caused by

several E coli strains harbouring virulence factors which are as yet unknown. Indeed,

recent genome-sequencing studies of various E coli strains have determined a core

genome of only 30% harboured by all these strains, making this possibility a

challenging concept.85

Lipopolysaccharide (LPS) endotoxins, present in all gram-negative bacteria, play a

major role in the etiology of CM (Hacker et al., 2004). Like bacteria, endotoxins enter

via the uterus, gut, and mammary gland. The systemic clinical signs elicited by

endotoxin release are complex, as various endogenous mediators are involved in

pathogenesis. The relevance of E coli endotoxins initiating complex reactions in the

animal organism has been proven before (Ramasoota et al., 2000; Magnusson et al.,

2001). The administration of coliform endotoxins via intravenous, intramammary,

intrauterine, or subcutaneous application causes clinical and blood chemical changes

similar to those in natural CM cases (Nachreiner and Ginther, 1969, 1974; Elmore et

al., 1978). For instance, subnormal serum concentrations of Ca++, Zn++, and iron is a

clear indication of endotoxin exposure (Holst et al., 1993), as is a rise in serum

cortisol levels (Magnusson et al., 1994). Furthermore, secretion of colostrum and milk

depends on the complex and well-balanced interaction of a series of different

hormones. These complex balances can be easily disturbed when LPS suppresses

the release of prolactin by the anterior pituitary, increasing cortisol concentrations

and decreasing circulating thyroid hormone (Smith and Wagner, 1985). Production

and secretion of milk are affected adversely by these changes.

Immune response and innate immunity

To a large extent, the outbreak of disease is determined by the interaction between

the invading microorganism and the host’s immune system. Clinical signs of CM are

most often seen in the first 24 hours after parturition, indicating a strong connection to

the postpartum period. In an experimental setting, Magnusson et al. (2001) found

that the time of inoculation of bacteria into the mammary gland influenced the

development of disease: clinical signs were seen in sows infected 48 hours, but not

96 hours, before parturition. Furthermore, the number of circulating

16

polymorphonuclear neutrophils was higher in sows that were more prone to develop

disease. Whether this fact can be related to the presence of other microorganisms

was not defined, but all sows had been diagnosed as healthy at the beginning of the

infection trial (Magnusson et al., 2001). Possibly, an exaggerated response to

bacterial infections, causing tissue injury, aggravates clinical signs (Magnusson et al.,

2001). Moreover, lysozyme, an enzyme that non-specifically stimulates the

phagocytic activity of leucocytes and the level of immunoglobulins, was present in

high concentrations in sows from herds of low CM prevalence (Wawron, 1995). After

experimental inoculation of E coli (0.5 mL of bacterial suspension per teat, 105 colony

forming units per mL), Österlundh et al. (2002) showed no significant differences in

functional capacities of granulocytes in sows affected and non-affected by CM.

After inoculation of 12 sows with E coli by the intramammary route (0.5 mL of

bacterial suspension per teat, 105 colony forming units per mL), Zhu et al. (2007a)

detected an increase in proinflammatory cytokines. The mammary glands appeared

capable of producing IL-1ß, IL-6, IL-8, and TNF-α, and the authors concluded that

local cytokine mRNA expression differs between mammary glands of sows that do or

do not develop clinical signs of mastitis. Especially TNF-α is considered to be a

useful indicator to monitor the severity and course of CM (Nakajima et al., 1997; Zhu

et al., 2004; Zhu et al., 2007b). Löving and Magnusson (2002) showed a significantly

higher density of CD4+ and CD8+ cells in animals developing clinical mastitis

compared to those without clinical disease, supporting the theory that massive

inflammatory reactions are triggered by endotoxins. In addition, in this study by

Löving and Magnusson (2002), sows developing clinical disease had a lower density

of MHC class II+ cells. This down-regulation can be related to the adverse effects of

LPS. Therefore, the authors postulated that the outcome of mammary infection was

related to sensitivity to LPS rather than to an ineffective immune response (Löving

and Magnusson, 2002).

Furthermore, the immune response is modified both by cortisol and oestrogen

affecting resistance to infection (Kelley et al., 1994), and both hormones vary

considerably in their concentration at the time of parturition. Resistance to infection in

swine is also influenced by sex hormones (Magnusson and Einarsson, 1990;

Magnusson and Fossum, 1992). However, in another study by Magnusson et al.

17

(2001), a difference in concentration of these hormones could not be identified in

sows with and without CM, suggesting that development of mastitis in sows before

parturition is not modulated by cortisol and oestrogen.

Treatment

After diagnosis of CM, antibiotic treatment must be started as soon as possible to

reduce the negative effects on both the sow and the piglets. Antibiotics are often

administered immediately after diagnosis to shorten the time period of undernutrition

for the piglets, but antimicrobial susceptibility is not tested. Therefore, the use of

broad spectrum antimicrobials administered parenterally, for example amoxicillin

(Markowska-Daniel and Kolodziejczyk, 2001), tylosin (Waldmann and Wendt, 2001),

or potentiated sulphonamides (Waldmann and Wendt, 2001), is indicated. Antibiotics

must reach effective levels in the mammary gland; consequently, pharmacokinetics

have to be considered. Another antibiotic showing a concentration in colostrum and

milk explicitly above the minimum inhibitory concentration is enrofloxacin (Oliel and

Bertschinger, 1990). In several studies, its use as a highly efficient antibiotic given

orally at 2.5 mg per kg body weight twice a day is recommended (Schöning and

Plonait, 1990; Rose et al., 1996; Scuka et al., 2006a). In a study on the therapeutic

performance of the cephalosporin cefquinom, this antibiotic, injected intramuscularly

at doses of 2 mg per kg body weight every 24 hours for 3 days was more efficient

than the control drug, amoxicillin (Heinritzi and Hagn, 1999).

In order to reduce inflammatory reactions, therapy with non-steroidal anti-

inflammatory drugs (NSAIDs), especially meloxicam at 0.4 mg per kg body weight

per sow in a single injection, has become popular in recent years (Hirsch et al., 2003;

Hoy and Friton, 2005). The advantages of this treatment are better recovery rates

and reduced piglet weight losses (Hoy and Friton, 2005). Use of flunixin meglumine

combined with enrofloxacin achieved no advantages compared to use of enrofloxacin

alone (Sterr, 2001). Occasionally, oxytocin (10 IU), injected five times at 2- to 3-hour

intervals, can initiate milk production (Waldmann and Wendt, 2001). However, as

routine use of oxytocin is associated with poorer herd performance (Ravel et al.,

1996), overuse should be avoided.

18

The effect of prostaglandin F2α (PGF2α) injection is controversial: in some herds, the

risk for periparturient disorders was minimized (Baer and Bilkei, 2005), while in

others, no effect could be proven (Hansen and Jacobsen, 1976; Ehnvall et al., 1977).

Prostaglandin F2α has its main impact on uterine debris postpartum, and, therefore,

administration in cases of CM is not indicated. As proposed by Kirkwood (Kirkwood,

1999), in the absence of vulval discharge problems, PGF2α does not improve sow

and litter performance. An alternative attempt to treat CM with bee venom was

proposed by Choi and Kang (Choi and Kang, 2001). Animals treated with apitherapy

showed significantly shorter periods of abnormal milk secretion (clots, blood traces,

or discoloration) compared to animals receiving antibiotic treatment with penicillin G

at 400,000 IU per animal. Besides treatment of sows, all economically reasonable

efforts to save the piglets should be attempted. To save the litter, piglets can be

cross-fostered or fed milk replacer (Klopfenstein et al., 2006).

From the very first recognition of CM as a problem in sows, there have been various

efforts to reduce prevalence of CM by a considerable number of measures. Nutrition

management is proposed as a useful tool to minimize the risk of CM (Persson et al.,

1989). High-fibre diets in late gestation have been used to decrease the occurrence

of early lactation problems, but it is unclear whether fibre addition or resultant protein

dilution in the feed ration is the cause of a lower prevalence of CM (Klopfenstein et

al., 2006). Feed reduction before parturition is a widespread practice and might

reduce not only obstipation, but also the amount of faeces produced. Consequently,

the exposure of the teats to contamination is reduced, and CM risk decreases as well

(Klopfenstein et al., 2006). On the day before and after farrowing, provision of ad

libitum drinking water is recommended (Waldmann and Wendt, 2001).

Supplementation with lactulose as a prebioticum in periparturient sows results in

better sow and piglet performance (Cosgrove and Foxcroft, 1996). Other measures to

avoid obstipation are feeding of linseed and other laxatives and adequate exercise

for the sow (Bilkei and Horn, 1991; Cosgrove and Foxcroft, 1996). Good hygiene

practice with all-in, all-out management, adequate temperatures in the farrowing

houses, and introduction of sows to clean farrowing houses 10 to 14 days prior to

farrowing are management factors that should be taken into account (Hammerl et al.,

1995). Manual interventions, eg, manual obstetrics in the peripartal period, should be

reduced to a minimum. Nevertheless, neither this nor other management practices

19

are able to totally prevent CM. Identification and reduction of risk factors, combined

with excellent hygiene management, are the only way to cope with a herd problem in

the long term (Hoy, 2003).

Non-specific paramunity inducers like an immunostimulator containing inactivated

Parapovirus ovis (Bayer AG, Leverkusen, Germany) were proved to have positive

effects on sows affected by CM (Choi and Kang, 2001). However, after natural

infection, mammary glands did not develop resistance to subsequent infections

(Bertschinger et al., 1990). Therefore, the effect of vaccines against E coli with regard

to CM can be doubted. Furthermore, there must be strict adherence to subcutaneous

injection of the vaccine, as the same dose administered via intramuscular or

intravenous injection may cause severe endotoxemia (Garcia et al., 1998;

Bertschinger, 1999). While vaccinations against infections with enterotoxigenic E coli

in piglets are commercially available and show positive effects (Haesebrouck et al.,

2004), the current knowledge about pathogen-host-interactions in CM is still too

limited to develop useful prevention tools.

Conclusion and future approaches

Commercial sow lines from pig breeding companies are continuously being improved

in their reproductive capabilities, with large litters and high-milk-producing potential,

and pigs are therefore exposed to a physiologically extreme situation during and

soon after birth. Although severe forms of CM are rare, piglet mortality and failure to

gain weight contribute to the outstanding economic relevance of this disease

complex. The demands for sufficient growth rate of suckling piglets and greater litter

size puts pressure on the lactating sow. The transition from gestation to lactation is of

paramount importance to sufficient milk yield and prevalence of CM during that

period. High piglet mortality, poor growth of suckling piglets, and poor average

weaning weights can be prevented only when CM is approached in a holistic way.

The current method to deal with postparturient disorders includes immediate

antibiotic treatment of sows if body temperature is above a defined threshold. This

threshold is defined rather subjectively and the use of it might be regarded critically,

since increases and decreases in body temperature may appear physiological. To

minimize the administration of antibiotics, it is therefore essential to diagnose CM and

PPDS not only by temperature increase, but also by a combination of appropriate

20

criteria. A threshold of 39.5°C in the time frame 12 to 24 hours postpartum is

recommended to avoid confusion of fever with physiological hyperthermia (Gerjets et

al., 2008).

Prevention is the best way to cope with CM in a population, but difficult to

accomplish, as the etiology of CM is extremely variable. At the current state of

knowledge, the reason for only some sows developing clinical signs of infection after

contact with ubiquitous bacteria remains unknown. The immune response and the

actual development of clinical signs seem to depend on the immunological reactivity

of the sow. Hence, one may hypothesize that developing clinical CM is largely

dependent on the individual resistance of the sow. Immune competence, including

resistance to infections, is genetically determined (Mallard et al., 1992; Magnusson

and Greko, 1998). The heritability for CM resistance is approximately 10% (Lingaas

and Ronningen, 1991; Berg et al., 2001). As shown by Heringstad et al. (1999; 2003)

for mastitis resistance in dairy cattle, it is possible to achieve a sustainable selection

response for disease traits of low heritability. Thus, the analysed heritabilities for CM

resistance indicate the opportunity to use this trait for selection (Lingaas and

Ronningen, 1991; Berg et al., 2001). In pig production, genetic disease resistance,

particularly resistance against certain E coli, is applied as a breeding tool in the

United States, Canada, Denmark, and Switzerland. Since infectious organisms

evolve resistance against drugs used to control them, as shown for pathogens that

cause CM (Acar and Rostel, 2001), and since the costs of treatment and veterinary

care are increasing faster than the value of animals, breeding for enhanced disease

resistance offers a number of advantages over other control measures. Additionally,

on the basis of the current knowledge of E coli strains involved in CM, no common

virulence factor has been identified. To discover this genetic component in the

involved E coli strains and other bacterial species is an immense challenge for further

research, as are the scientific questions relating to CM in general.

21

References

Acar, J., Rostel, B., 2001. Antimicrobial resistance: an overview. Revue Scientifique

Et Technique De L Office International Des Epizooties 20, 797-810.

Armstrong, C.H., Hooper, B.E., Martin, C.E., 1968. Microflora associated with

agalactia syndrome of sows. Am. J. Vet. Res. 29, 1401-1407.

Awad Masalmeh, M., Baumgartner, W., Passering, A., Silber, R., Hinterdorfer, F.,

1990. Bakteriologische Untersuchungen bei an puerperaler Mastitis (MMA-

Syndrom) erkrankten Sauen verschiedener Tierbestände Österreichs

(Bacteriological studies in sows with pueperal mastitis in different herds in

Austria). Tierarztl. Umsch. 45, 526-535.

Bäckström, L., Morkoc, A.C., Connor, J., Larson, R., Price, W., 1984. Clinical study of

mastitis-metritis-agalactia in sows in Illinois. J. Am. Vet. Med. Assoc. 185, 70-

73.

Baer, C., Bilkei, G., 2005. Ultrasonographic and gross pathological findings in the

mammary glands of weaned sows having suffered recidiving mastitis metritis

agalactia. Reprod. Dom. Anim. 40, 544-547.

Berg, P., Andersen, S., Henryon, M., Nielsen, J., 2001. Genetic variation for birth

assistance and MMA in sows and diarrhoea in their litters. 52nd Annual

Meeting of the European Association for Animal Production, Budapest,

Hungary.

Berner, H., 1971. [Significance of chronic urinary tract diseases in the pathogenesis

of puerperal diseases and mastitides of sows]. Dtsch. Tierarztl. Wochenschr.

78, 241-245.

Bertschinger, H.U., 1999. Escherichia coli infections. In: Straw, B.E., D'Allaire, S.,

Mengeling, W.L., Taylor, D.J. (Eds.), Diseases of Swine. Iowa State University

Press, Ames, pp. 431-468.

Bertschinger, H.U., Bürgi, E., Eng, V., Wegmann, P., 1990. Lowering of the incidence

of puerperal mastitis in the sow by protection of the mammae from

contamination. Schweiz. Arch. Tierheilkd. 132, 557-566.

Bertschinger, H.U., Pohlenz, J., Hemlep, I., 1977a. Mastitis metritis agalactia

syndrome (milk fever) in sows. II. Bacteriological findings in spontaneous

cases. Schweiz. Arch. Tierheilkd. 119, 223-233.

Bertschinger, H.U., Pohlenz, J., Middleton Williams, D.M., 1977b. Investigations on

the mastitis metritis agalactia syndrome (milk fever) in sows. III. Induction of

22

klebsiella mastitis by teat inoculation. Schweiz. Arch. Tierheilkd. 119, 265-

275.

Bilkei, G., Bolcskei, A., Clavadetscher, E., Goos, T., Hofmann, C., Bilkei, H., Szenci,

O., 1994a. The periparturient disease complex of the sow .1. Evaluation of the

course of the bacteriuria of urine samples of old sows during the periparturient

period. Berliner Munchener Tierarztl. Wochenschr.107, 327-330.

Bilkei, G., Bölcskei, A., Goos, T., Hofmann, C., Szenci, O., 1994b. Die Beteiligung

von E. coli an puerperalen Urogenitaltraktinfektionen bei Altsauen (The

prevalence of Escherichia coli in urogenital tract infections of sows with a

previous history of MMA, PHS and SUGD). Tierarztl. Umsch. 49, 471-476.

Bilkei, G., Horn, A., 1991. The therapy of the metritis, mastitis, agalactia (MMA)

complex of swine. Berliner Munchener Tierarztl. Wochenschr. 104, 421-423.

Bostedt, H., Maier, G., Herfen, K., Hospen, R., 1998a. Clinical examinations on gilts

with pueperal septicaemia and toxaemia. Tierarztl. Prax. 26, 332-338.

Bostedt, H., Maier, G., Herfen, K., Hospen, R., 1998b. Clinical examinations on gilts

with puerperal septicaemia and toxaemia. Tierarztl. Prax. 26, 332-338.

Choi, S.-H., Kang, S.-S., 2001. Therapeutic effect of bee venom in sows with

hypogalactia syndrome postpartum. J Vet Sci 2, 121-124.

Cosgrove, J.R., Foxcroft, G.R., 1996. Nutrition and reproduction in the pig: Ovarian

aetiology. Anim. Reprod. Sci. 42, 131-141.

Ehnvall, R., Einarsson, S., Larsson, K., Segerstad, C.H.A., Westerberg, L., 1977.

Prostaglandin-induced parturition in swine - field-study on its accuracy after

treatment with different amounts of Pgf2alpha. Nordisk. Veterinaer. Medicin.

29, 376-380.

Elmore, R.G., Martin, C.E., 1980. Agalactia in current therapy. In: Morrow, D.A.

Theriogenology 2: Diagnosis, treatment and prevention of reproductive

diseases in animals. Philadelphia; WB Saunders Co, 114-121

Elmore, R.G., Martin, C.E., Berg, P., 1978. Absorption of Escherichia coli endotoxin

from the mammary glands and uteri of early postpartum sows and gilts.

Theriogenology 10, 439-445.

Furniss, S.J., 1987. Measurement of rectal temperature to predict mastitis, metritis

and alagactia (MMA) in Sows after farrowing. Prev. Vet. Med. 5, 133-139.

23

Garcia, P., Holst, H., Magnusson, U., Kindahl, H., 1998. Endotoxin-effects of

vaccination with Escherichia coli vaccines in the pig. Acta. Vet. Scand. 39,

135-140.

Gerjets, I., Kruse, S., Krieter, J., Kemper, N., 2008. Diagnosis of MMA affected sows:

bacteriological differentiation, temperature measurement and water intake.

Proc Int Vet Pig Soc Congr. Durban, South Africa.

Gooneratne, A.D., Hartmann, P.E., Nottage, H.M., 1982. The initiation of lactation in

sows and the mastitis-metritis-agalactia syndrome. Anim. Reprod. Sci. 5, 135-

140.

Grün, D., Reiner, G., Dzapo, V., 1993. Untersuchungen über Rassenunterschiede in

der Milchleistung beim Schwein. II. Mitteilung: Milchinhaltsstoffe und ihr

Einfluss auf die Entwicklung der Ferkel in der Säugephase (Investigations on

breed differences in milk yield of swine. II. Commitment: milk composition and

ist relation towards piglet development during suckling period). Repr. Dom.

Anim. 28, 22-27.

Hacker, J., Hochhut, B., Middendorf, B., Schneider, G., Buchrieser, C., Gottschalk,

G., Dobrindt, 2004. Pathogenomics of mobile genetic elements of toxigenic

bacteria. Int. J. Med. Microbiol. 293(7-8), 453-461.

Haesebrouck, F., Pasmans, F., Chiers, K., Maes, D., Ducatelle, R., Decostere, A.,

2004. Efficacy of vaccines against bacterial diseases in swine: what can we

expect? Vet. Microbiol. 100, 255-268.

Hammerl, J., Wolf, G., Berner, H., 1995. Klinische Untersuchungen zur Wirkung des

Paramunitätsinducers Baypamun als Prophylaxe beim MMA-Komplex der

Sau (Clinical study on the efficacy of the paramunity-inducer Baypamun as

prophylactic treatment of the MMA-syndrome in sows). Tierarztl. Umsch. 50,

383-386.

Hansen, L.H., Jacobsen, M.J., 1976. Incidence of MMA after prostaglandin-induced

parturition. Nordisk. Veterinaer. Medicin. 28, 357-360.

Hartmann, P.E., Smith, N.A., Thompson, M.J., Wakeford, C.M., Arthur, P.G., 1997.

The lactation cycle in the sow: physiological and management contradictions.

Livest. Sci. 50, 75-87.

Heinritzi, K., Gindele, H.R., Gerald, R., Schnurrbusch, U., 2006.

Schweinekrankheiten (Diseases in pigs). Stuttgart, Verlag Eugen Ulmer.

24

Heinritzi, K., Hagn, J., 1999. Comparison of therapeutic performance of the new

Cephalosporin Cefquinome with other treatment regimes in gilts with puerperal

septicaemia and toxaemia syndrome. Tierarztl. Prax. 27, 114-121.

Hellbrugge, B., Tolle, K.H., Bennewitz, J., Henze, C., Presuhn, U., Krieter, J., 2008.

Genetic aspects regarding piglet losses and the maternal behaviour of sows.

Part 1. Genetic analysis of piglet mortality and fertility traits in pigs. Animal 2,

1273-1280.

Heringstad, B., Klemetsdal, G., Ruane, J., 1999. Short communication: Validation of

estimates of genetic trend in the Norwegian cattle population. J. Dairy. Sci. 82,

2811-2812.

Heringstad, B., Klemetsdal, G., Steine, T., 2003. Selection responses for clinical

mastitis and protein yield in two Norwegian Dairy Cattle selection experiments.

J. Dairy. Sci. 86, 2990-2999.

Hermannson, I., Einarsson, S., Larsson, K., Bäckström, L., 1978. On the agalactia

postpartum in the sows: a clinical study. Nor. Vet. Med. 30, 465-473.

Hirsch, A.C., Philipp, H., Kleemann, R., 2003. Investigation on the efficacy of

meloxicam in sows with mastitis-metritis-agalactia syndrome. J. Vet.

Pharmacol. Ther. 26, 355-360.

Hirsch, A.C., Philipp, H.C., Kleemann, R., 2004. Efficacy of meloxicam for treatment

of the mastitis-metritis-agalactia syndrome in sows. Prakt. Tierarzt 85, 842-

848.

Hogan, J.S., Smith, K.L., 1997. Bacteria counts in sawdust bedding. J. Dairy. Sci. 80,

1600-1605.

Holst, H., Edquist, L.E., Kindahl, H., 1993. Reduced response to intravenous

endotoxin injections following repeated orl administration of endotoxin in the

pig. Acta. Vet. Scand. 34, 405-419.

Hoy, S., 2002. Investigations on influence of different housing factors on frequency of

puerperal diseases in sows. Prakt. Tierarzt. 83, 990-996.

Hoy, S., 2003. Auswirkungen der Puerperalerkrankungen bei Sauen auf die

Fruchtbarkeitsleistung (Investigations on the effects of pueperal diseases in

sows on the fertility). Arch. Tierz. 46 341-346.

Hoy, S., Friton, G.M., 2005. Effects of meloxicam treatment on performance of sows

with MMA syndrome. Prakt. Tierarzt. 86, 568-576.

25

Hühn, U., Rehbock, F., 2008. Prostaglandine contra Umrauschen. Krankhaftem

Scheidenausfluss nach der Geburt Paroli bieten. (Prostaglandines versus

open sows: stand up to fight postparturient discharge.). Dlz agrarmagazin 2,

1273-1280.

Hurley, W.L., 2001. Mammary gland growth in the lactating sow. Livest. Sci. 70, 149-

157.

Jones, J.E.T., 1979. Acute coliform mastitis in the sow. Annales de médecine

vétérinaire 19, 97-101.

Jorsal, S.E., 1983. Epidemiological and statistical analysis of disease records from

pig testing stations. Copenhagen.

Kelley, K.W., Johnson, R.W., Dantzer, R., 1994. Immunology discovers physiology.

Vet. Immunol. Immunopathol. 43, 157-165.

Kirkwood, R.N., 1999. Influence of postpartum cloprostenol injection on sow and litter

performance. Swine Health Prod 7, 121-122.

Kiss, D., Bilkei, G., 2005. A new periparturient disease in Eastern Europe,

Clostridium difficile causes postparturient sow losses. Theriogenology 63, 17-

23.

Klopfenstein, C., Farmer, C., Martineau, G.P., 2006. Diseases of the Mammary

Glands and Lactation Problems. In: Straw, B.E., Zimmermann, J.J., Taylor,

D.J. (Eds.), Diseases of swine. Iowa State University Press, pp. 833-860.

Kobera, R., 2000. Vergleichende Prüfung der klinischen Wirksamkeit von Cefquinom

und Enrofloxacin bei der Behandlung des Mastitis-Metritis-Agalaktie-Komlexes

der Sau (Comparative analysis of the efficacy of Cefquinime and

Enrofloxacine in the treatment of the MMA-complex in the sow) (doctoral

thesis). Universität Leipzig, Leipzig.

Krieter, J., Presuhn, U., 2005. Genetic variation for MMA treatment in sows. 57th

EAAP,. Book of abstracts, p. 62.

Krüger, M., Schroedl, W., Isik, K., Lange, W., Hagemann, L., 2002. Effects of

lactulose on the intestinal microflora of periparturient sows and their piglets.

Eur J Nutr 41, I/26-I/31.

Kuhn, N.J., 1969. Progesterone withdrawal as the lactogneic trigger in rat. J.

Endocrinol. 44, 39-54.

Larsen, I., Thorup, F., 2006. The diagnosis of MMA. Proc Int Vet Pig Soc Congr,

Copenhagen, Denmark .

26

Lingaas, F., Ronningen, K., 1991. Epidemiological and genetical studies in

Norwegian pig herds. V. Estimates of heritability and phenotypic correlations

of the most common diseases in Norwegian pigs. Acta. Vet. Scand. 32, 115-

122.

Liptrap, R.M., 1980. Prostaglandin F2alpha and progesterone in experimental

hypogalactia in sows. Res. Vet. Sci. 29, 240-247.

Löving, M., Magnusson, U., 2002. Sows intramammarily inoculated with Escherichia

coli at parturition: II. Effects on the densities of MHC class II+, CD4+ and

CD8+cells in the mammary gland. Vet. Immunol. Immunopathol. 90, 45-54.

Madec, F., Leon, E., 1992. Farrowing Disorders in the Sow - a Field-Study. J. Vet.

Med. Ser. A-Physiol. Pathol. Clin. Med. 39, 433-444.

Magnusson, U., Einarsson, S., 1990. Effects of exogenous oestradiol on the number

and functional capacity of circulating mononuclear and polymorphonuclear

leukocytes in the sow. Vet. Immunol. Immunopathol. 25, 235-247.

Magnusson, U., Fossum, C., 1992. Effect of estradiol-17ß treatment of gilts on blood

mononuclear cell functions in vitro. Am. J. Vet. Res. 53, 1427-1430.

Magnusson, U., Greko, C., 1998. Assessment of phagocytic capacity and opsonic

activity in blood and mammary secretion during lactation in sows. J. Vet. Med.

B 45, 353-361.

Magnusson, U., Holst, H., Kindahl, H., Karlsson, A., 1994. Effect of mimicking

prepartum concentrations of estradiol-17beta on the inflammatory response to

endotoxins in gilts. Am. J. Vet. Res. 55, 785-789.

Magnusson, U., Mörner, A.P., Persson, A., Karlstam, E., Sternberg, S., Kindahl, H.,

2001. Sows intramammarily inoculated with Escherichia coli: Influence of time

of infection, hormone concentrations and leucocyte numbers on development

of disease. J. Vet. Med., Series B 48, 501-512.

Mallard, B.A., Wilkie, B.N., Kennedy, B.W., Quinton, M., 1992. Use of estimated

breeding values in a selection index to breed Yorkshire pigs for high and low

immune and innate resistance factors. Anim. Biotech. 3, 257-280.

Markowska-Daniel, I., Kolodziejczyk, P., 2001. The usefulness of Amoksiklav susp. in

the control of mastitis-metritis-agalactia syndrome in sows. Slov. Vet. Res. 38,

333-342.

Martin, C.E., 1970. Status of the mastitis-metritis-agalactia syndrome of sows. J. Am.

Vet. Med. Assoc. 157, 1519-1521.

27

Martin, C.E., Hartmann, P.E., Gooneratne, A.D., 1978. Progesterone and

corticosteroides in the initiation of lactation in the sow. Aust. J. Biol. Sci. 31,

517-525.

Martin, C.E., Hooper, B.E., Armstrong, C.H., 1967. A clinical and pathological study

of the mastitis-metritis-agalactia syndrome of sows. J. Am. Vet. Med. Assoc.

151, 1629-1634.

Mauch, C., Bilkei, G., 2004. The influence of prepartal bacteriuria on the reproductive

performance of the sow. Dtsch, Tierarztl. Wochenschr. 111, 166-169.

Mirko, C.P., Bilkei, G., 2004. Acute phase proteins, serum cortisol and preweaning

litter performance in sows suffering from periparturient disease. Acta. Vet.

Scand. 54, 153-161.

Morkoc, A., Backstrom, L., Lund, L., Smith, A.R., 1983. Bacterial endotoxin in blood

of dysgalactic sows in relation to microbial status of uterus, milk, and intestine.

J. Am. Vet. Med. Assoc. 183, 786-789.

Nachreiner, R.F., Ginther, O.J., 1969. Current studies on the mastitis-metritis-

agalactia complex of swine in Wisconsin. J. Am. Vet. Med. Assoc. 155, 1860-

1865.

Nachreiner, R.F., Ginther, O.J., 1974. Induction of agalactia by administration of

endotoxin (Escherichia coli) in swine. Am. J. Vet. Res. 35, 619-622.

Nachreiner, R.F., Ginther, O.J., Rebelin, W.E., Carlson, I.H., 1971. Pathological and

endocrinologic changes associated with porcine agalactia. Am. J. Vet. Res.

35, 619-622.

Nakajima, Y., Mikami, O., Yoshioka, M., Motoi, Y., Ito, T., Ishikawa, Y., Fuse, M.,

Nakano, K., Yasukawa, K., 1997. Elevated levels of tumor necrosis factor-

alpha (TNF-alpha) and interleukin-6 (IL-6) activities in the sera and milk of

cows with naturally occurring coliform mastitis. Res. Vet. Sci. 62, 297-298.

Oliel, N., Bertschinger, H.U., 1990. Prophylaxis of experimentally induced coliform

mastitis in the sow with enrofloxacin (Baytril). Proc Int Vet Pig Soc Congr,

Lausanne, Switzerland.

Österlundh, I., Holst, H., Magnusson, U., 1998. Hormonal and immunological

changes in blood and mammary secretion in the sow at parturition.


Österlundh, I., Hulten, F., Johannisson, A., Magnusson, U., 2002. Sows intrammarily

inoculated with Escherichia coli at parturition: I. Functional capacity of

28

granulocytes in sows affected or non-affected by clinical mastitis. Vet.

Immunol. Immunopathol. 90, 35-44.

Papadopoulos, G.A., Janssens, G.P.J., Dewulf, J., Vanderhaeghe, C., Maes, D.G.D.,

2008. Risk factors implicated in the occurrence of post-partum dysgalactia

sydrome in sows. Proc Int Vet Pig Soc Congr, Durban, South Africa.

Papadopoulos, G.A., Maes, D.G.D., Van Weyenberg, S., Verheyen, A., Janssens,

G.P.J., 2007. Selected parameters in urine as indicators of milk production in

lactating sows: A pilot study. Vet J; In Press, Corrected Proof.

Pedersen Mörner, A., Faris, A., Krovacek, K., 1998. Virulence Determinants of

Escherichia coli Isolated from the Milk of Sows with Coliform Mastitis. J. Vet.

Med., Series B 45, 287-295.

Penny, R.H.C., 1970. The agalactia complex in the sow: a review. Aust. Vet. J. 46,

153-159.

Persson, A., 1997. Clinical assessment of udder health status of sows at time of

weaning with special reference to bacteriology and cytology in milk.

Zentralblatt fur Veterinärmedizin. Reihe A 44, 143-158.

Persson, A., Pedersen, A.E., Goransson, L., Kuhl, W., 1989. A long-term study on

the health-status and performance of sows on different feed allowances during

late pregnancy .1. Clinical observations, with special reference to agalactia

post-partum. Acta. Vet. Scand. 30, 9-17.

Persson, A., Pedersen Morner, A., Kuhl, W., 1996a. A long-term study on the health

status and performance of sows on different feed allowances during late

pregnancy. II. The total cell content and its percentage of polymorphonuclear

leucocytes in pathogen-free colostrum and milk collected from clinically

healthy sows. Acta. Vet. Scand. 37, 279-291.

Persson, A., Pedersen Morner, A., Kuhl, W., 1996b. A long-term study on the health

status and performance of sows on different feed allowances during late

pregnancy. III. Escherichia coli and other bacteria, total cell content,

polymorphonuclear leucocytes and pH in colostrum and milk during the first 3

weeks of lactation. Acta. Vet. Scand. 37, 293-313.

Petersen, B., 1983. Methods of early recognition of puerperal and fertility disorders in

the sow. Livest. Sci. 10, 253-264.

29

Plonait, H., Bickhart, K., 1997. Geburt, Puerperium und perinatale Verluste. In:

Plonait, H., Bickhart, K. (Eds.), Lehrbuch der Schweinekrankheiten. Parey,

Berlin/Hamburg, pp. 472-512.

Prunier, A., de Braganca, M.M., Le Dividich, J., 1997. Influence of high ambient

temperature on performance of reproductive sows. Livest. Sci. 52, 123-133.

Ramasoota, P., Krovacek, K., Chansiripornchai, N., Morner, A.P., Svenson, S.B.,

2000. Identification of Escherichia coli recovered from milk of sows with

coliform mastitis by random amplified polymorphic DNA (RAPD) using

standardized reagents. Acta. Vet. Scand. 41, 249-259.

Ravel, A., DAllaire, S., BigrasPoulin, M., 1996. Influence of management, housing

and personality of the stockperson on preweaning performances on

independent and integrated swine farms in Quebec. Prev Vet Med 29, 37-57.

Ringarp, N., 1960. Clinical and experimental investigations into a post-parturient

syndrome with agalactia in sows. Acta. Agric. Scand. Suppl 7.

Ringmar-Cederberg, E., Johnson, L., 1996. Sow culling in Sweden. Report 111, Proc.

NFJ-seminar no. 265. Longevitiy of sows, Foulum, Denmark.

Rooke, J.A., Bland, I.M., 2002. The acquisition of passive immunity in the new-born

piglet. Livest. Sci. 78, 13-23.

Rose, M., Schnurrbusch, U., Heinritzki, K., 1996. The use of cefquinomein in the

treatment of pig respiratory disease and the MMA syndrome. Proc Int Vet Pig

Soc Congr, Bologna, Italy.

Ross, R.F., Harmon, R.L., Zimmermann, B.J., Young, T.F., 1983. Susceptibility of

sows to experimentally induced Escherichia coli mastitis. Am. J. Vet. Res. 44,

949-954.

Ross, R.F., Orning, A.P., Woods, R.D., Zimmermann, B.J., Cox, D.F., Harris, D.L.,

1981. Bacteriologic study of sow agalactia. Am. J. Vet. Res. 42, 949-955.

Schöning, G., Plonait, H., 1990. Metaphylaxe und Therapie des MMA-Syndromes bei

Sauen mit Baytril (Metaphylaxis and therapy of the MMA-syndrome in sows

with Baytril). Dtsch. Tierarztl. Wochenschr. 97, 5-10.

Scuka, L., Stukelj, M., Valencak, Z., 2006a. Therapeutic effects of enrofloxacin in

mastitis-metritis-agalactia syndrome: a review. Acta. Vet. Brno. 75, 5151-5522.

Smith, B.B., Martineau, G.P., Bisaillon, A., 1992. Mammary glands and lactation

problems. In: Leman, A.D., Straw, B.E. Mengeling, W.L., D’Allaire, S., Taylor,

D.J.(Eds.), Diseases of swine. Iowa State University Press, pp. 523-634.

30

Smith, B.B., Wagner, W.C., 1985. Effect of Escherichia coli endotoxin and

thyrotropin-releasing hormone on prolactin in lactating sows. Am. J. Vet. Res.

46, 175-180.

Sterr, K., 2001. Untersuchungen zur Wirksamkeit von Flunixin-Meglumin in der

Therapie des Mastitis-Metritis-Agalaktie-Syndroms der Sauen (Study on the

efficacy of Flunixin-Meglumin in a therapy regime against the mastitis-metritis-

agalactia syndrome in sows) (doctoral thesis). Universität München,

Tierärztliche Fakultät.

Sujatha, K., Srilatha, C.H., Nisar Ahmed, N., 2003. Metritis Mastitis Agalctia

syndrome in sows. Ind. Vet. J. 80, 379-380.

Swarbrick, O., 1968. The porcine agalactia syndrome. Clinical and histological

observations. Vet. Rec. 82, 241-252.

ten Napel, J., Kemp, B., Luiting, P., de Vries, A.G., 1995. A biological approach to

examine genetic variation in weaning-to-oestrus interval in first-litter sows.

Livest. Sci. 41, 81-93.

Thorup, F., 2000. Effect of treatment for MMA-retrospective observations. Proc Int

Vet Pig Soc Congr, Melbourne, Australia.

Waldmann, K.-H., 2000. Behandeln Sie MMA-Probleme rechtzeitig. top agrar 3, 4-7.

Waldmann, K.-H., Wendt, M., 2001. Lehrbuch der Schweinekrankheiten (Handbook

of pig diseases). Parey Verlag, Stuttgart.

Waller, C.M., Bilkei, G., Cameron, R.D.A., 2002. Effect of periparturient diseases

accompanied by excessive vulval discharge and weaning to mating interval on

sow reproductive performance. Aust. Vet. J. 80, 545-549.

Wawron, W., 1995. An investigation of non-specific immunity in sows suffering from

the mastitis-metritis-agalactia-syndrome. Tierarztl. Umsch. 50, 584-590.

Wegmann, P., Bertschinger, H.U., 1984. Sequential cytological and bacteriological

examination of the secretions from sucked and unsucked mammary glands

with and without mastitis. Proc Int Vet Pig Soc Congr, Mexico City, Mexico.

Wegmann, P., Bertschinger, H.U., H., J., 1986. A field study on the prevalence of

coliform mastitis (MMA syndrome) in Switzerland and the antimicrobial

susceptibility of the coliform bacteria from the milk. Proc Int Vet Pig Soc

Congr, Barcelona, Spain.

31

Zhu, Y., Berg, M., Fossum, C., Magnusson, U., 2007a. Proinflammatory cytokine

mRNA expression in mammary tissue of sows following intramammary

inoculation with Escherichia coli. Vet. Immunol. Immunopathol. 116, 98-103.

Zhu, Y., Fossum, C., Berg, M., Magnusson, U., 2007b. Morphometric analysis of

proinflammatory cytokines in mammary glands of sows suggests an

association between clinical mastitis and local production of IL-1beta, IL-6 and

TNF-alpha. Vet. Res. 38, 871-882.

Zhu, Y., Österlundh, I., Hulten, F., Magnusson, U., 2004. Tumor necrosis factor

alpha, interleukin-6, serum amyloid A, haptoglobin, and cortisol concentrations

in sows following intramammary inoculation of Escherichia coli. Am. J. Vet.

Res. 65, 1434-1439.

32

Chapter 2

Comparison of virulence gene profiles of Escherichia coli isolates from

sows with Coliform mastitis and healthy sows

Imke Gerjets, M. sc. agr.*1

Imke Traulsen, Dr. sc. agr. 1

Kerstin Reiners, Dr. sc. agr. 2

Nicole Kemper, Prof. Dr. med. vet. 3


24098 Kiel, Germany

2PIC Germany GmbH, Ratsteich 31, D-24837 Schleswig Germany



Accepted for publication in Veterinary Microbiology

33

Abstract

Coliform mastitis (CM) is not only a serious economical and animal welfare touching

problem in dairy cattle, but also in sows after farrowing. Due to this disease, the

essential adequate supply with colostrum for the growth and the health of the piglets

is not ensured. Besides other influencing factors, Escherichia (E.) coli is of great

importance as a causative agent of this multifactorial disease. In this study, E. coli

isolates from milk samples of healthy and CM-affected sows were examined for the

presence of virulence genes associated with extraintestinal E. coli strains,

enterotoxigenic E. coli and other pathogenic E. coli.

The isolated E. coli harbored mainly virulence genes of extraintestinal E. coli strains

(especially fimC, ompA, traT, hra, kpsMTII, iroN). The virulence gene spectrum for

both samples from CM-affected and healthy sows did not differ significantly.

Particular virulence gene profiles of E. coli isolates from diseased sows were not

detected.

This study provides novel insights into the role of E. coli in association with mastitis in

sows since it is the first time E. coli isolates from CM-affected sows’ milk were

analysed for virulence genes. Because there were no differences in the prevalence of

E. coli and their virulence-associated genes between healthy and diseased sows,

other causative factors seem to have greater influence on the pathogenesis of

porcine CM.

Keywords

ETEC, ExPEC, multiplex PCR, swine, virulence factors

34

Introduction

‘Coliform mastitis’ (CM) is the main symptom of puerperal disorders occurring in sows

after farrowing which are subsumed under the term postpartum dysgalactia

syndrome (PPDS or PDS) (Gerjets and Kemper, 2009; Klopfenstein et al., 2006). The

etiology of CM is multifactorial with husbandry, management, feeding and hygiene as

influencing factors (Klopfenstein et al., 2006), but mainly bacteria are the causative

agents for the inflammation. In bacteriological analyses, especially Escherichia (E.)

coli was isolated, but the strains were not further investigated for virulence-

associated genes. Strains of E. coli can be broadly classified into three groups by

their location and their characteristic virulence genes: commensal E. coli, intestinal

pathogenic E. coli (IPEC) colonizing the intestine, and extraintestinal pathogenic E.

coli (ExPEC) that reach extraintestinal niches like the urinary tract (Russo and

Johnson, 2000). In swine, especially enterotoxigenic E. coli (ETEC) as a pathotype of

IPEC are well described as causal agents for severe diseases like diarrhea in

neonatal and weaned piglets (Casey and Bosworth, 2009). The ExPEC pathotypes

are e.g. causative for urinary tract infections (uropathogenic E. coli (UPEC)) or

septicaemia in pigs (Daigle et al., 1997; Krag et al., 2009; Shpigel et al., 2008). A

selection of virulence genes known to be associated with ETEC, ExPEC pathotypes

and shiga toxin-producing E. coli (STEC) is listed in Table 1.

A new putative pathotype of ExPEC was proposed by Shpigel et al. (2008):

mammary pathogenic E. coli, with as specific set of virulence genes, which are

associated with mastitis in dairy animals. However, up to now epidemiological studies

have not shown a common virulence gene profile for these E. coli so far (Kaipainen

et al., 2002; Srinivasan et al., 2007; Wenz et al., 2006).

The aim of our study was to analyse the occurrence of different virulence genes in E.

coli isolates associated with Coliform mastitis in sows.

35

Table 1: Prevalence of virulence-associated genes in E. coli isolates from healthy/diseased sows (*P<0.05)

Gene(s)/categories prevalence of virulence-associated genes (%) P-value E. coli isolates (n = 1,271)

of CM-negative sows E. coli isolates (n = 1,132)

of CM-positive sows Adhesins afa / dra ExPEC - - - fimC ExPEC 82.30 84.72 0.1112 hra* ExPEC 11.33 14.84 0.0106 iha ExPEC 0.16 0.18 0.9077 sfa / foc ExPEC 0.08 0.18 0.4971 K99 (fanA) ETEC - - - K88 (faeG) ETEC 0.08 0.09 0.9367 987P (fasA) ETEC 0.08 - 0.3443 F18 (fedA) ETEC - 0.09 0.2892 F41 (fedA subunit) ETEC - - -

Iron chuA* ExPEC 4.80 6.71 0.0434 iron* ExPEC 9.28 12.37 0.0148 sitD chr. ExPEC 0.24 0.62 0.1461 sitD ep. ExPEC 5.74 6.27 0.5858

Protectins neuC ExPEC 0.39 0.18 0.3251 kpsMT II* ExPEC 9.99 13.07 0.0178 ompA ExPEC 37.61 35.34 0.2480 traT ExPEC 49.80 52.12 0.2568

Toxins hlyA ExPEC 1.65 2.56 0.1189

Enterotoxins STII ETEC - 0.18 0.1338 STI ETEC 2.28 1.94 0.5658 LT ETEC - 0.09 0.2892

Shiga Toxins Stx2e STEC - - -

Invasins gimB ExPEC 0.08 0.00 0.3452 ibeA ExPEC 0.63 0.97 0.3443

Miscellaneous pic ExPEC 0.63 1.33 0.0804 malX ( RPai) ExPEC - 0.18 0.1338

Materials and methods

Animals and study design

The investigation was carried out between April 2008 and August 2010 on five

multiplication herds in Germany (A - E), supervised by PIC Germany GmbH

36

Schleswig (Table 2). The farms were of high health status and tested free from

enzootic pneumonia, rhinitis, Actinobacillus pleuropneumoniae and dysentery. The

number of sows housed in the farms ranged from 700 to 1,800. The sows were in

different parities (1–9) and of different lines (Landrace, Large White and crossbreds,

partly with Duroc).

They were identified as CM-affected when their rectal temperature was above 39.5°C

24 h post partum (Furniss, 1987) and the mammary glands showed symptoms of

inflammation. In addition, the appearance and the performance of the piglets were

evaluated with regard to their behavior and body condition. Healthy half- or full-sib

sows from the same farrowing group that farrowed closest in time served as controls.

The half-sib design was chosen due to further studies on the genetic background via

genotyping (Preißler et al., unpublished data). In total, 2,005 milk samples were

examined (1,026 milk samples from sows with CM and 979 from healthy sows).

Before gathering a collective sample of several teats, mammary glands were cleaned

and disinfected with disinfection swabs containing 70% isopropyl-alcohol. The first

streams of milk were discarded whereas the followings were milked on transport

swabs with Amies medium (transwab, medical wire & equipment, Corsham,

England). The milk samples were stored at 4°C before sending them to the

laboratory within 72 hours.

Table 2: Number of milk samples and E. coli isolates of five different farms

Farm number of milk samples number of E. coli isolates CM-negative CM-positive CM-negative CM-positive

A 498 501 600 477 B 13 15 16 21 C 276 323 460 481 D 25 20 32 27 E 167 167 163 126

total 979 1,026 1,271 1,132

Bacteriological analysis

The swabs were incubated in Caso broth for 24 h at 37°C. With a plastic loop, 10 µL

of the enrichment were streaked onto Columbia blood agar and Endo agar (both

Oxoid, Cambridge, United Kingdom) and incubated aerobically another 24 h at 37°C.

The grown bacteria were differentiated by their morphology, haemolysis on blood

agar and Gram staining. Pure cultures were grown on blood agar after another 24 h

37

incubation at 37°C before biochemical confirmation to species level with the

identification system API (bioMérieux, Craponne, France).

Escherichia coli isolates were distinguished due to individual morphology on blood

agar and API 20E. All isolates were selected for further investigations.

Desoxyribonucleic acid of the identified E. coli strains was prepared by solving a few

colonies in 200 µL distilled water. After boiling for 10 min and centrifugation, 3 µL of

the supernatant was taken for PCR analysis. The presence of virulence genes

associated with ExPEC strains, ETEC and other pathogenic E. coli was determined

by multiplex PCR (mPCR) assays for all E. coli isolates, as described by Ewers et al.

(2007) and Casey and Bosworth (2009).

In total, 2,403 isolates were tested for the presence of 27 virulence genes for the

following virulence factors (Table 1): heat labile toxin (LT), heat stable toxin a and b

(STI, STII), Shiga toxin (Stx2e), capsular polysaccharide (neuC), group II capsule

antigen (kpsMTII), outer membrane protein (ompA), transfer protein (traT), heme

receptor gene (chuA), catecholate siderophore receptor (iroN), iron transport system

genes (sitD chr., sitD ep.), haemolysin A (hlyA), invasins (gimB, ibeA), serin protease

autotransporter (pic), different adhesins and fimbrial genes (afa/draB, fimC, hra, iha,

sfa/foCD, K88, K99, 987P, F41, F18) and pathogenicity-associated island marker

(RPai (malX)). Controls for molecular assays were avian pathogenic E. coli (APEC)

strain IMT2470, UPEC strains IMT7920 and IMT9267 and ETEC strains IMT204,

IMT19, IMT4830 and IMT3838 (Casey and Bosworth, 2009; Ewers et al., 2007),

kindly provided by the Institute of Microbiology and Epizootics of the Free University

Berlin.

Statistical analysis

The statistical analysis was performed using the procedures FREQ and CORR from

the Statistical Analysis System (SAS Institute Inc., 2005). Chi square-tests were used

to analyse differences in virulence gene frequencies between diseased and healthy

sows. Statistical significance was indicated in two levels: P<0.05* and P<0.01**.

Pearson correlation coefficients, calculated to show associations between virulence

genes, were presented as heatmaps. Heatmaps representing gene prevalence were

generated to allow assessment of the virulence genes regarding occurrence and

distribution (R Development Core Team, 2009). Virulence genes in the heatmaps

were arranged automatically according to their means. Genes with similar means are

ordered close together.

38

Correlations and heatmaps were performed with only those virulence genes detected

in more than 1 % of the analysed E. coli strains.

Results

Escherichia coli strains

Escherichia coli was found in 70.6 % (n=724) of the milk samples of CM-affected and

in 77.9 % (n=762) of the milk samples of non-infected sows. In total, 1,132 E.coli

isolates from CM-positive samples and 1,271 isolates from CM-negative samples

were identified and further examined by mPCR. The median number of isolates in

milk samples of both diseased and healthy sows was one (Figure 1).

Of the 2,403 E. coli isolates, 593 harbored one virulence gene, 983 two, 357 three

and 369 four or more virulence genes. In 101 E. coli isolates, no virulence-associated

genes were found. The E. coli isolates from CM-positive as well as from negative

sows had a median number of two virulence genes.

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

0 1 2 3 4

number of E. coli strains per milk sample

fre

qu

en

cy

in

%

samples of CM-positive sows (n=1,026) samples of CM-negative sows (n=979)

Figure 1: Number of E. coli isolates in milk samples from CM-positive and -negative sows

Comparison of virulence gene profiles

A variety of virulence genes was identified consisting of mainly those associated with

ExPEC (98.9 % of E. coli isolates from diseased and 99.0 % of E. coli isolates from

healthy sows) (Table 1). The highest prevalence was found for the type 1 fimbriae

fimC (in 84.7 % of the isolates of diseased and 82.3 % of the isolates of healthy

39

sows) and for the protectins ompA and traT (in 35.3 % and 52.1 % of the isolates

from CM-positive, and 37.6 % and 49.8 % of the isolates from CM-negative sows,

respectively). Other genes identified in 9.3 to 14.8 % of the E. coli isolates were hra,

kpsMTII and iroN. Almost all of the virulence-associated factors were more often

detected in E. coli isolates of CM-affected sows than in isolates of healthy sows,

except 987P, neuC, ompA and gimB.

The virulence genes hra, chuA, iroN and kpsMTII occurred significantly more

frequently in isolates of diseased animals. The same applied for particular

combinations of these genes (Table 3), except for the profiles chuA - iroN, kpsMTII –

chuA - iroN and kpsMTII – hra – chuA - iroN. Those combinations were also less

prevalent in all E. coli isolates. The greatest difference between diseased and healthy

sows was found for the virulence gene profile chuA - hra (2.7 % in E. coli of CM-

positive and 0.9 % in E. coli of CM-negative sampled sows, respectively). In total,

there were no obvious patterns specific for either diseased or healthy sows.

Table 3: Prevalence of virulence gene profiles in E. coli isolates from clinically CM-diseased and healthy sows (*P<0.05, **P<0.01)

Virulence gene profile prevalence of E. coli isolates (%) with respective gene profile from:

P-value

samples (n = 979) of CM-negative sows

samples (n = 1,026) of CM-positive sows

kpsMTII* 9.99 13.07 0.0178 chuA* 4.80 6.71 0.0434 hra* 11.33 14.84 0.0106 iron* 9.28 12.37 0.0148 kpsMTII, chuA** 0.87 2.21 0.0068 kpsMTII, hra** 2.44 4.42 0.0073 kpsMTII, iron** 1.42 3.09 0.0052 chuA, hra** 0.94 2.65 0.0014 chuA, iron 0.24 0.44 0.3823 hra, iron* 1.34 2.65 0.0204 kpsMTII, chuA, hra* 0.63 1.59 0.0231 kpsMTII, chuA, iron 0.08 0.27 0.2634 kpsMTII, hra, Iron** 0.31 1.33 0.0052 chuA, hra, Iron* - 0.35 0.0339 kpsMTII, chuA, hra, iroN - 0.27 0.0663

40

Correlations between virulence genes

Statistical analysis of associations between all virulence factors of the E. coli isolates

is shown in Figure 2. Several similar patterns in the heatmaps were visible for

virulence genes of strains from CM-positive and negative sows: the gene hlyA is

positively associated with chuA and pic; iroN is positively associated with ompA and

sitDepi, respectively. Highest positive correlations existed between the genes iroN

and sitDepi for both isolates from diseased and healthy sows. The genes traT and

fimC were also highly positive correlated, but only in E. coli isolates of CM-negative

sows.

41

chuA traT

iroN

ompA

sitD

epi

kpsM

TII

hlyA pi

c

fimC

hrA

ST

I

CM-positive sows

STI

hrA

fimC

pic

hlyA

kpsMTII

sitDepi

ompA

iroN

traT

chuA

chuA traT

iroN

ompA

sitD

epi

kpsM

TII

hlyA pi

c

fimC

hrA

ST

I

CM-negative sows

STI

hrA

fimC

pic

hlyA

kpsMTII

sitDepi

ompA

iroN

traT

chuA

Figure 2: Statistical associations between 12 virulence-associated genes from E. coli isolates of CM-positive and CM-negative sows. Colours range from light grey (little associated) to dark grey (highly associated) (p<0.05). Gaping spaces indicate no significant correlation between virulence genes.

42

Gene prevalence with regard to different seasons and farms

The gene traT was more often found in E. coli isolates of samples of CM-positive

sows in winter whereas STI (heat stable toxin a) was only found in summer (Figure

3). The gene chuA occurred more frequently in E. coli isolates of positive sows in

winter and autumn and iroN in summer, autumn and winter as well as kpsMTII was

always more prevalent in samples of diseased sows. All virulence genes were found

more often in E. coli isolates of diseased sows in all seasons except for ompA and

traT which were more prevalent in isolates of healthy sows in spring.

However, the differences in occurrence of the genes were greater between the

seasons than between CM-positive and negative sows.

The same held true for the influence of the farms on the occurrence of virulence

genes. The gene STI was only found in E. coli isolates sampled from farm A whereas

kpsMTII was more prevalent in samples from farm D. The gene traT occurred more

often in isolates from diseased sows. The gene prevalence on the farms differed only

slightly between CM-infected and healthy sows. Differences regarding the occurrence

of the mentioned virulence genes in the seasons and farms were significant (P<0.05).

43

Figure 3: Heatmaps representing gene prevalence in E. coli isolates (n=2,403) of different CM-status (neg, pos), seasons (spring, summer, autumn, winter) and farms (A, B, C, D, E). Colours range from light grey (gene found in 1 - 5 % of the isolates) to dark grey (gene found in 80 - 88 % of the isolates).

44

Discussion

The aim of the study was to analyse and compare virulence genes of E. coli isolates

from milk samples of CM-positive and CM-negative sows, because virulence gene

profiles of E. coli isolates associated with mastitis has not been described so far

(Kaipainen et al., 2002; Srinivasan et al., 2007; Wenz et al., 2006). Escherichia coli is

the pathogen most frequently isolated in association with porcine puerperal disorders

(Armstrong et al., 1968; Awad Masalmeh et al., 1990; Bertschinger et al., 1977a;

Ross et al., 1981). It was also isolated in high frequencies in milk samples of

diseased sows in this investigation as well as in milk from healthy sows.

The detailed analysis of virulence-associated genes of the E. coli isolates revealed

only slight differences between isolates of diseased and healthy sows (P<0.05).

Although there were single genes or gene combinations with a greater linkage to E.

coli isolates from milk samples of CM-affected sows, there were no specific virulence

gene patterns detectable. Heatmaps were performed to allow a visualization of

correlations among virulence genes of isolates of different CM-status, seasons and

farms.

The E. coli strains were isolated using an enrichment of the milk samples. This

qualitative culture procedure was used to promote the growth of the E. coli strains, as

described before for faecal samples (Hussain et al., 2010; Wu et al., 2010).

Regarding the actual presence of virulence genes, an influence of enrichment

procedures has only been described in detail for STEC (Vimont et al. 2007), but has

not been proven for incubation in Caso-Broth for the applied duration. However, a

possible influence on the quantitative proportion of different strains cannot be

excluded though faecal contamination of the samples was minimized by a strict

sampling protocol.

Escherichia coli strains causing acute coliform mastitis in dairy cattle originate from

the animal’s faecally contaminated environment and infect the udder via the teat

canal (Eberhart, 1984). Experiments by Bertschinger et al. (1990) and Bertschinger

et al. (1977b), where the mammary glands of sows were protected against faecal

contamination, support the theory of a galactogenous route of infection via the teat

duct. Like bovine mastitis, porcine mastitis may also resemble urinary tract infection

as the infection may be ascending (Kaipainen et al., 2002). Among others, causative

agents of urinary tract infections (UTI) are UPEC, a pathotype of ExPEC. In contrast

to commensal E. coli isolates, UPEC harbor more virulence genes encoding virulent

45

capsule antigens, iron acquisition systems, adhesions and secreted toxins (Wiles et

al., 2008). The virulence genes iroN and fimC are reported as urovirulence factors

(Russo et al., 2002; Wiles et al., 2008) and were also identified in high percentages in

our study. In a survey by Won et al. (2009), the presence of 19 virulence-associated

genes in avian pathogenic E. coli (APEC), another pathotype of ExPEC, was

determined, and approximately 95 % of the APEC isolates possessed fimC.

However, fimC has also been frequently detected in non-pathogenic E. coli and is

proposed to be not highly associated with the pathogenesis of APEC-infections

(Kawano et al., 2006). We also found the fimbrial gene fimC in high prevalence in

isolates of healthy sows, confirming this theory.

The traT gene, detected in half of the examined E. coli strains, was found in milk of

CM-affected dairy cattle, too. Out of 160 Finnish isolates from cows with mastitis, 37

%, and out of 113 Israeli isolates, 41 % harbored traT (Kaipainen et al., 2002).

Nemeth et al. (1991) identified the gene in 43 % of E. coli strains isolated from the

milk of cows with mastitis. In another study by Acik et al. (2004), milk samples from

healthy cows and sheep were analysed and traT was present in 62.3 % of all isolates

(62.5 % of the isolates from cows and 60 % of the isolates from sheep).

All in all, a spectrum of virulence genes was present in bovine mastitis strains of E.

coli, but those strains do not possess specific virulence factors contributing to clinical

disease. Serum resistance was the only virulence property of E. coli consistently

associated with isolates of coliform mastitis in dairy cattle (Barrow and Hill, 1989;

Fang and Pyorala, 1996). A relationship between traT and serum resistance,

however, could not be confirmed (Nemeth et al., 1991; Vandekerchove et al., 2005).

The results and conclusions concerning the virulence genes related to bovine

mastitis are comparable to the findings of our study in sows. Specific sow factors,

e.g. the individual disposition of the animal, are probably more important and the host

defense status is generally accepted as key factor determining the outcome of the

disease (Burvenich et al., 2003). Current investigations deal with the genetic

background of CM via genotyping of diseased and healthy sows (Preißler et al.,

unpublished data).

In conclusion, a variety of virulence genes was detected among the E. coli isolates

for both samples from CM-positive and negative sows. The identified virulence genes

belonged mainly to the large group of genes related to ExPEC, but a categorization

into the pathotype ExPEC only by virulence gene typing was not possible. Many

46

virulence-associated factors (e.g. for iron-uptake systems, fimbriae and other

adhesions) are fitness factors which help the bacteria to adapt to and successfully

colonize the host so that the line between virulence and fitness properties of E. coli

strains is very thin (Dobrindt, 2005).

The results of our study support the hypothesis that any given E. coli strain, even

those considered to be non-pathogenic, can cause coliform mastitis in sows, if further

adversely environmental, genetic or other influencing factors promoting infection are

present.

Acknowledgements

This research project was funded by the German Federal Ministry of Education and

Research (BMBF) in the research programme ‘”FUGATO – Functional Genome

Analysis in Animal Organisms,” project “geMMA – structural and functional analysis

of the genetic variation of the MMA-syndrome” (FKZ0315138).

Prof. Dr. Lothar Wieler, Dr. Christa Ewer and Ines Diehl from the Institute of

Microbiology and Epizootics of the Free University Berlin are greatly acknowledged

for their technical support with the mPCR assays and for providing the control strains.

We also wish to thank Jens Wolfmueller and Evelyn Lass for technical assistance

and especially all farmers involved for their help with the sample acquisition.

References

Acik, M.N., Yurdakul, N.E., Cakici, L., Onat, N., Dogan, O., Cetinkaya, B., 2004. traT

and CNF2 genes of Escherichia coli isolated from milk of healthy cows and

sheep. Res. Vet. Sci. 77, 17-21.







Austria). Tierarzt. Umsch. 45, 526-535.

Barrow, P.A., Hill, A.W., 1989. The virulence characteristics of strains of Escherichia

coli isolated from cases of bovine mastitis in England and Wales. Vet.

Microbiol. 20, 35-48.

47

Bertschinger, H.U., Bürgi, E., Eng, V., Wegmann, P., 1990. Lowering of the incidence

of puerperal mastitis in the sow by protection of the mammae from

contamination. Schweiz. Arch. Tierheilkd. 132, 557-566.

Bertschinger, H.U., Pohlenz, J., Hemlep, I., 1977a. Mastitis metritis agalactia



Bertschinger, H.U., Pohlenz, J., Middleton Williams, D.M., 1977b. Investigations on

the mastitis metritis agalactia syndrome (milk fever) in sows. III. Induction of

Klebsiella mastitis by teat inoculation. Schweiz. Arch. Tierheilkd. 119, 265-275.

Burvenich, C., Van Merris, V., Mehrzad, J., Diez-Fraile, A., Duchateau, L., 2003.

Severity of E. coli mastitis is mainly determined by cow factors. Vet. Res. 34,

521-564.

Casey, T.A., Bosworth, B.T., 2009. Design and evaluation of a multiplex polymerase

chain reaction assay for the simultaneous identification of genes for nine

different virulence factors associated with Escherichia coli that cause diarrhea

and edema disease in swine. J. Vet. Diagn. Invest. 21, 25-30.

Daigle, F., Dozois, C.M., Jacques, M., Harel, J., 1997. Mutations in the f165(1)A and

f165(1)E fimbrial genes and regulation of their expression in an Escherichia

coli strain causing septicemia in pigs. Microb. Pathog. 22, 247-252.

Dobrindt, U., 2005. (Patho-)genomics of Escherichia coli. Int. J. M. Microbiol. 295,

357-371.

Eberhart, R.J., 1984. Coliform mastitis. Vet. Clin. North. Am. 6, 287-301.

Ewers, C., Li, G.W., Wilking, H., Kiessling, S., Alt, K., Antao, E.M., Laturnus, C.,

Diehl, I., Glodde, S., Homeier, T., Bohnke, U., Steinruck, H., Philipp, H.C.,

Wieler, L.H., 2007. Avian pathogenic, uropathogenic, and newborn meningitis-

causing Escherichia coli: How closely related are they? Int. J. M. Microbiol.

297, 163-176.

Fang, W., Pyorala, S., 1996. Mastitis-causing Escherichia coli: Serum sensitivity and

susceptibility to selected antibacterials in milk. J. Dairy Sci. 79, 76-82.


and agalactia (MMA) in sows after farrowing. Prev. Vet. Med. 5, 133-139.

Gerjets, I., Kemper, N., 2009. Coliform mastitis in sows: A review. J. Swine. Health

Prod. 17, 97-105.

48

Hussain, M.I., Wani, S.A., Nagamani, K., Fayaz, I., Mir, M.A., Nabi, A., Qureshi, K.,

Khan, M.A., Nishikawa, Y. 2010. Isolation and characterization of

enteroaggregative, enterotoxigenic, diffusely adherent Escherichia coli and

Salmonella Worthington from human diarrhoeic faecal samples in Kashmir and

Secundarbad, India. Wolrd J. Microbiol. Biotechnol. 26, 1883-1889.

Kaipainen, T., Pohjanvirta, T., Shpigel, N.Y., Shwimmer, A., Pyorala, S., Pelkonen,

S., 2002. Virulence factors of Escherichia coli isolated from bovine clinical

mastitis. Vet. Microbiol. 85, 37-46.

Kawano, M., Yaguchi, K., Osawa, R., 2006. Genotypic analyses of Escherichia coli

isolated from chickens with colibacillosis and apparently healthy chickens in

Japan. Microbiol. Immunol. 50, 961-966.

Klopfenstein, C., Farmer, C., Martineau, G.P., 2006. Diseases of the mammary

glands and lactation problems. In: Straw, B.E., Zimmermann, J.J., Taylor, D.J.

(Eds.), Diseases of swine, Iowa State University Press, Ames, pp. 833-860.

Krag, L., Hancock, V., Aalbaek, B., Klemm, P., 2009. Genotypic and phenotypic

characterisation of Escherichia coli strains associated with porcine

pyelonephritis. Vet. Microbiol. 134, 318-326.

Nemeth, J., Muckle, C.A., Lo, R.Y.C., 1991. Serum resistance and the traT gene in

bovine mastitis-causing Escherichia coli. Vet. Microbiol. 28, 343-351.

R, Development Core Team, 2009. R - A language and environment for statistical

computing. R - Foundation for Statistical Computing. Vienna Austria.

http://www.r-project.org/.



Russo, T.A., Johnson, J.R., 2000. Proposal for a new inclusive designation for

extraintestinal pathogenic isolates of Escherichia coli: ExPEC. J. Infect. Dis.

181, 1753-1754.

Russo, T.A., McFadden, C.D., Carlino-MacDonald, U.B., Beanan, J.M., Barnard, T.J.,

Johnson, J.R., 2002. IroN functions as a siderophore receptor and is a

urovirulence factor in an extraintestinal pathogenic isolate of Escherichia coli.

Infect. Immun. 70, 7156-7160.

SAS, 2005. Version 9.1, SAS Institute, Cary, NC, USA.

Shpigel, N.Y., Elazar, S., Rosenshine, I., 2008. Mammary pathogenic Escherichia

coli. Curr. Opin. Microbiol. 11, 60-65.

49

Srinivasan, V., Gillespie, B.E., Lewis, M.J., Nguyen, L.T., Headrick, S.I., Schukken,

Y.H., Oliver, S.P., 2007. Phenotypic and genotypic antimicrobial resistance

patterns of Escherichia coli isolated from dairy cows with mastitis. Vet.

Microbiol. 124, 319-328.

Vandekerchove, D., Vandemaele, F., Adriaensen, C., Zaleska, A., Hernalsteens,

J.P., De Baets, L., Butaye, P., Van Immerseel, F., Wattiau, P., Laevens, H.,

Mast, J., Goddeeris, B., Pasmans, F., 2005. Virulence-associated traits in

avian Escherichia coli: Comparison between isolates from colibacillosis-

affected and clinically healthy layer flocks. Vet. Microbiol. 108, 75-87.

Vimont, A., Vernozy-Rozand, C., Montet, M.P., Bavai, C., Fremaux, B., Delignette-

Muller, M.-L., 2007. Growth of shiga-toxin producing Escherichia coli (STEC)

and bovine feces background microflora in various enrichment protocols. Vet.

Microbiol. 123, 274-281.

Wenz, J.R., Barrington, G.M., Garry, F.B., Ellis, R.P., Magnuson, R.J., 2006.

Escherichia coli isolates' serotypes, genotypes, and virulence genes and

clinical coliform mastitis severity. J. Dairy Sci. 89, 3408-3412.

Wiles, T.J., Kulesus, R.R., Mulvey, M.A., 2008. Origins and virulence mechanisms of

uropathogenic Escherichia coli. Exp. Mol. Pathol. 85, 11-19.

Won, G.Y., Moon, B.M., Oh, I.G., Matsuda, K., Chaudhari, A.A., Hur, J., Eo, S.K., Yu,

I.J., Lee, Y.J., Lee, Y.S., Kim, B.S., Lee, J.H., 2009. Profiles of virulence-

associated genes of avian pathogenic Escherichia coli isolates from chickens

with colibacillosis. J. Poult. Sci. 46, 260-266.

Wu, Y., Hinenoya, A., Taguci, T., Nagita, A., Shima, K., Tsukamoto, T., Sugimoto, N.,

Asakura, M., Yamasaki, S. 2010. Distribution of virulence genes related to

adhesins and toxins in shiga toxin-producing Escherichia coli strains isolated

from healthy cattle and diarrheal patients in Japan. J. Poult. Sci. 46, 260-266.

50

Chapter 3

Assessing individual sow risk factors for coliform mastitis in sows:

A case-control study


Imke Traulsen, Dr. sc. agr.1

Kerstin Reiners, Dr. sc. agr.2

Nicole Kemper, Prof. Dr. med. vet.3


24098 Kiel, Germany

2 PIC Germany GmbH, Ratsteich 31, D-24837 Schleswig Germany



Preventive Veterinary Medicine (2011), doi:10.1016/j.prevetmed.2011.04.012

51

Abstract

In order to investigate sow-specific risk factors associated with coliform mastitis, a

case-control study was performed over the course of 28 months. Data of three farms

were collected under production conditions. Sows suffering from coliform mastitis

after farrowing served as cases, and healthy half- or full-sib sows from the same farm

served as controls. Individual sow characteristics and the seasonal influence were

analysed by conditional logistic regression. The final multivariate model identified four

risk factors: the risk of suffering from coliform mastitis increased with a higher number

of piglets born alive and stillborn piglets. Gilts had an increased risk for the disease,

and birth intervention was also associated with a higher prevalence of mastitis. Birth

induction and season had no significant influence on the occurrence of coliform

mastitis.

The time during and soon after farrowing is a very sensitive period in pig production

demanding great attention by the farmer. With respect to the economic losses,

monitoring of potentially endangered sows as well as detailed documentation and

selection of disease cases are of particular importance when coping with coliform

mastitis.

Keywords

Birth intervention, fever, litter size, mastitis, post-parturient sow

52

Introduction

Coliform mastitis (CM) is an economically relevant disease in sows (Bertschinger and

Fairbrother, 2006). The average prevalence in herds is about 13 %, but also

prevalence up to 60 % has been reported (Bäckström et al., 1984; Hirsch et al., 2003;

Krieter and Presuhn, 2009). After farrowing, the infection of the mammary gland

results in reduced productivity of the sows and increased mortality of the piglets. The

affected animals suffer from fever and an inflammation of the glands mostly followed

by decreased milk secretion 24 to 48 hours post partum. Therefore, the sows fail to

meet the needs of their piglets. A detailed description of the disease is given in a

recent review (Gerjets and Kemper, 2009).

Coliform mastitis is a multifactorial disease, i.e. several factors influence the

prevalence of mastitis among herds. The term ‘coliform mastitis’ refers to a clinical

mastitis due to coliform bacteria (Escherichia species (spp.), Klebsiella spp.,

Enterobacter spp. and Citrobacter spp.) which have been found to be associated with

the disease complex in many studies (Awad Masalmeh et al., 1990; Bertschinger and

Fairbrother, 2006; Hirsch et al., 2003; Ross et al., 1981).

Most studies concerning the identification of risk factors were carried out between

1970 and 1990. Potential factors were mostly related to housing, management and

feeding practices, and tested in univariate analyses. Changes in housing (Waldmann

and Wendt, 2001), single housing and lack of exercise (Ringarp, 1960; Hoy, 2002) as

well as overfeeding in late gestation (Göransson, 1989) are only some factors

reported to increase the occurrence of CM.

A previous study by Papadopoulos et al. (2010) dealt with management and strategy-

related risk factors, acquired via questionnaires, in a multivariable analysis, but

focused on the herd level.

The main objective of this study was to identify potential risk factors for CM, in

particular individual sow characteristics related to production parameters, for CM by

performing a case-control study.

53

Material and methods

Data collection and study design

Data were collected from three multiplication herds in Germany, supervised by PIC

Germany GmbH Schleswig, from April 2008 to August 2010 within the scope of a

microbiological study (Gerjets et al., 2010, submitted).

The farms were chosen because of their similar high health status and the available

documented reproduction data. They were tested free from porcine reproductive and

respiratory syndrome-virus, rhinitis, Actinobacillus pleuropneumoniae dysentery and

enzootic pneumonia. The number of sows housed on the farms varied between

1,000 and 1,800. The sows were of different parities (1 to 9) and lines (Landrace,

Large White and crossbreds, partly with Duroc). Information about cross-fostering

was not documented.

All sows were examined after farrowing and considered as CM cases when their

rectal temperature was above the threshold of 39.5°C 24 h post partum (Furniss,

1987) and the mammary glands showed definite signs of inflammation.

Healthy half- or full-sib sows from the same herd and, if possible, the same farrowing

group served as controls. The half-sib design was chosen due to further studies on

the genetic background via genotyping (Preißler et al., unpublished data).

In total, data of 1.337 sows were analysed (683 CM-affected and 654 healthy sows).

The investigation was carried out as m:n matched case-control study, i.e. one or

more cases were matched to one or more controls due to their respective herd and

relationship. Therefore, the number of cases and controls within one group varied

from 1 to 57 and 1 to 62, respectively.

The case-control study investigated factors associated with CM, especially

reproduction parameters of the sows. Information of the ‘sire’, the ‘number of piglets

born alive’ and ‘stillborn piglets’, the ‘parity number’, ‘birth induction’, ‘birth

intervention’ and the ‘season’ was recorded for the clinical cases and for the controls.

Statistical analyses

The unit of analysis was the sow. The dependent variable was the occurrence of CM

as a binary outcome (present or absent). Independent variables (potential risk

factors) were categorised by checking the distribution of the observations. A χ2-test

was calculated pairwise to determine whether independent variables were correlated.

The relation between potential risk factors and the occurrence of CM was analysed

by conditional logistic regression with the procedure LOGISTIC (SAS Institute Inc.,

54

2005), in which the sire and the farm served as strata. First, risk factors were tested

in univariate analysis. Those that were associated with the outcome variable at

P<0.25 were then included in the multivariate analysis. P-values for the variables

were based on the Wald statistic. The Akaike information criterion and the Schwartz

criterion evaluated the goodness of fit of the final models.

Results

Depending on the CM status, the average number of piglets born alive of the

analysed sows varied between 11.9 and 12.5, and the average number of stillborn

piglets ranged from 0.9 to 1.6. The mean number of weaned piglets varied between

10.1 and 10.6 (Table 1).

Table 1: Means (standard deviations) of reproduction traits of the investigated sows

Variable Case-control study Cases (n=683) Controls (n=654)

No. of piglets born alive 12.1 (3.1) 11.9 (3.0) No. of stillborn piglets 1.1 (1.6) 0.9 (1.3) No. of weaned piglets 10.5 (2.1) 10.5 (1.9)

An overview of the levels of the collected animal-specific parameters and their

respective number of cases and controls is given in Table 2.

Results of univariate and multivariate analyses of the case-control study are shown in

Table 2 and 3. The risk factors ‘number of piglets born alive’ and ‘stillborn piglets’, the

‘parity number’ and ‘birth intervention’ were associated with the occurrence of CM.

The parameters ‘birth induction’ and ‘season’ did not influence the occurrence of CM

significantly and were not included in the multivariate analysis.

The chance of suffering CM significantly increased when the number of piglets born

alive was higher than 13 (OR = 1.65). The Odds Ratio was also higher when there

were more than one stillborn piglets (OR = 1.45). Primiparous sows had a twofold

higher chance of contracting the disease than older ones. Birth intervention

increased the chance of suffering CM (OR = 1.72).

55

Table 2: Levels and univariate conditional logistic regression analyses* of potential risk factors for CM in sows (n = 1337). OR = odds ratio; CI = 95% confidence interval; P = P-value.

Variable Level Case-control study Univariate

No. of cases / %

(n=683)

No. of controls / %

(n=654)

OR

95% CI

P

No. of piglets born alive

< 12

12-13 > 13

257 (37.6)

195 (28.6) 231 (33.8)

263 (40.2)

194 (29.7) 197 (30.1)

1.00 1.01a 1.40b

- 0.75-1.35 1.04-1.87

0.0460

No. of stillborn piglets 0

1 > 1

318 (46.5)

165 (24.2) 200 (29.3)

340 (52.0)

160 (24.4) 154 (23.6)

1.00 1.09a 1.45b

- 1.07-1.96 0.81-1.46

0.0540

No. of parity 1

2-3 4-5 > 5

156 (22.8)

265 (38.8) 142 (20.8) 120 (17.6)

98 (15.0)

287 (43.9) 173 (26.5) 96 (14.6)

1.00 0.64a 0.60a 0.76b

- 0.44-0.92 0.38-0.95 0.43-1.36

0.0630

Birth intervention No

Yes

507 (74.3)

175 (25.7)

547 (83.6)

107 (16.4)

1.00 1.62

- 1.19-2.20

0.0020

Birth induction No

Yes

274 (40.1)

409 (59.9)

278 (42.5)

376 (57.5)

1.00 0.96

- 0.74-1.25

0.7726

Season Spring

Summer Autumn Winter

183 (26.7)

226 (33.1) 182 (26.7) 92 (13.5)

169 (25.8)

222 (34.0) 168 (25.7) 95 (14.5)

1.00 0.86a 0.85a 0.96a

- 0.63-1.19 0.60-1.21 0.63-1.40

0.7604

* Matched on farm and sire a.b Different letters within an effect show significant differences between categories

56

Table 3: Multivariate conditional logistic regression analyses* of potential risk factors for CM in sows (n =

1337). OR = odds ratio; CI = 95% confidence interval; P = P-value.

Variable/ Levels Multivariate OR 95% CI P

No. of piglets born alive < 12 1.00 - 0.0032 12-13 1.07a 0.80-1.45 > 13 1.65b 1.21-2.24 No. of stillborn piglets 0 1.00 - 0.0595 1 1.04a 0.77-1.40 > 1 1.45b 1.06-1.99 No. of parity 1 1.00 - 0.0146 2-3 0.59a 0.40-0.85 4-5 0.51a 0.32-0.81 > 5 0.67b 0.37-1.21 Birth intervention No 1.00 - 0.0008 Yes 1.72 1.25-2.37 * Matched on farm and sire a.b Different letters within an effect show significant differences between categories

Discussion

The aim of the study was to identify potential sow-specific risk factors via a matched

case-control study. Data were collected from three herds, possibly limiting the

generalizability of the study, but allowing to focus on the single animal. Environmental

influences were standardised through the recording of data on these three farms with

their high health standards. According to the microbiological and genetic study

design, and in contrast to previous investigations, the emphasis was put on analysing

the individual sow. Milk samples of cases and controls were taken for bacteriological

analysis, confirming Escherichia coli as main pathogen associated with CM (Gerjets

et al., unpublished data).

A higher number of piglets born alive was associated with a higher risk for the sows

of becoming diseased. This is in accordance with findings of Bostedt et al. (1998), in

which gilts with 1.1 piglets more than healthy sows suffered significantly more often

from feverish puerperal illness, and also showed an increased stillbirth rate.

Concerning the number of stillborn piglets, our study also supports these results.

However, other researchers did not find differences in the number of stillborn piglets

between diseased and healthy sows (Mirko and Bilkei, 2004; Van Gelder and Bilkei,

2005).

57

Literature about the effect of the parity number on the occurrence of mastitis is

contradictory. While Baer and Bilkei (2005) found sows of higher parity (>4) having

an increased risk of suffering mastitis, other studies have described a greater mastitis

risk for lower parity sows (1. and 2. parity) (Bostedt et al., 1998; Hoy, 2003; Krieter

and Presuhn, 2009). We also found a higher risk for primiparous sows, leading to the

interpretation that those sows were more prone to disease. Explanations for this

might be their not fully developed immune system (Wendt, 2000; Hoy, 2006), or that

sows suffering mastitis in their first parity might be culled. Physiological hyperthermia

is also often observed in postparturient sows, especially primiparous ones, leading to

misinterpretations (Klopfenstein et al., 2006; Gerjets et al., 2008).

However, most of the sow reproduction parameters identified are interrelated

naturally, as shown by Bostedt et al. (1998).

According to the available documented parameters of this study, practical

recommendations for the prevention of CM by management measures are limited.

The litter size, the rate of stillborn piglets and the parity may give hints when dealing

with CM. For the single sow, these parameters, obtained around the time of farrowing

and therefore the time of possible CM, are less informative for the current litter.

However, considered over following parities and on herd level, these data stress the

need for a precise documentation and monitoring.

Commercial sow lines from pig breeding companies are continuously improved in

their reproductive capabilities, with large litters and high-milk-producing potential. The

sows are therefore exposed to a physiologically extreme situation during and soon

after birth. Prevention is the best way to cope with CM in a population, but difficult to

accomplish, as the etiology of CM is extremely variable. The investigated factor birth

intervention may be helpful in order to prevent CM in sows because it is associated

with a higher risk for mastitis and can be regarded in the management. This fact has

also been reported by Bostedt et al. (1998). Papadopoulos et al. (2010) pointed out

that frequent supervision of farrowing by the stockpersons may reduce the incidence

of postpartum dysgalactia syndrome. These contradictory findings could be explained

by different interpretations of the terms birth intervention and supervision. Frequent

supervision might be positive because all factors prolonging the duration of the birth

process increase the prevalence of CM (Berner, 1971; Petersen, 1983; Bostedt et al.,

1998) and therefore, supervision, resulting in a reduced birth duration, decreases the

risk of CM. On the contrary, manual intervention leading to a manipulation of the birth

58

process might have a negative influence, especially if accompanied by insufficient

hygiene.

Conclusions

The identification of potential risk factors is the key element in preventing diseases.

We found that the risk of CM in sows increased with a higher number of piglets born

alive and stillborn piglets, with a lower parity and the application of birth intervention.

These results should be taken into account when coping with problem herds, e.g. by

the detailed documentation of disease on a single animal basis and the careful

selection of non-diseased sows for further production cycles.

Acknowledgements


Research (BMBF) in the research programme “FUGATO – Functional Genome



References








73.



agalactia. Reprod. Domest. Anim. 40, 544-547.

59

Berner, H., 1971. Significance of chronic urinary tract diseases in the pathogenesis of

puerperal diseases and mastitides of sows. Deut. Tierarztl. Woch. 78, 241-

245.

Bertschinger, H.U., Fairbrother, J.M., 2006. Escherichia coli infections. In: Straw,

B.E., D'Allaire, S., Mengeling, W.L., Taylor, D.J. (Eds.), Diseases of swine,

Iowa State University Press, Ames, pp. 431-468.

Bostedt, H., Maier, G., Herfen, K., Hospen, R., 1998. Clinical examinations on gilts



and agalactia (MMA) in sows after farrowing. Prev. Vet. Med. 5, 133-139.

Gerjets, I., Kemper, N., 2009. Coliform mastitis in sows: A review. J. Swine. Health

Prod. 17, 97-105.




Gerjets, I., Traulsen, I., Reiners, K., Kemper, N., 2010. Comparison of virulence gene

profiles of Escherichia coli isolates from sows with coliform mastitis and

healthy sows. Submitted.

Göransson, L., 1989. The effect of feed allowance in late pregnancy on the

occurence of agalactia post partum in the sow. J. Am. Vet Med Assoc. 36,

505-513.






Hoy, S., 2003. Auswirkungen der Puerperalerkrankungen bei Sauen auf die

Fruchtbarkeitsleistung (Investigations on the effects of puerperal diseases in

sows on the fertility). Arch. Tierz. 46 341-346.

Hoy, S., 2006. The impact of puerperal diseases in sows on their fertility and health

up to next farrowing. Anim. Sci. J. 82, 701-704.


Glands and Lactation Problems, in: Straw, B.E., Zimmermann, J.J., Taylor,


60

Krieter, J., Presuhn, U., 2009. Genetic variation for MMA treatment. Zuchtungskunde

81, 149-154.



Scand. 54, 153-161.

Papadopoulos, G.A., Vanderhaeghe, C., Janssens, G.P.J., Dewulf, J., Maes, D.G.D.,

2010. Risk factors associated with postpartum dysgalactia syndrome in sows.

Vet. J. 184, 167-171.

Petersen, B., 1983. Methods of early recognition of puerperal and fertility disorders in

the sow. Livest. Sci. 10, 253-264.

Ringarp, N., 1960. A post-parturient syndrome with agalactia in sows. Acta. Agric.

Scand. Suppl. 7, 1-166.



SAS, 2005. Version 9.1, SAS Institute, Cary, NC, USA.

Van Gelder, K.N., Bilkei, G., 2005. The course of acute-phase proteins and serum

cortisol in mastitis metritis agalactia (MMA) of the sow and sow performance.

Tijdschr. Diergeneeskd. 130, 38-41.

Waldmann, K.-H., Wendt, M., 2001. Lehrbuch der Schweinekrankheiten. Parey

Verlag, Stuttgart.

Wendt, M., 2000. So optimieren Sie das Geburtsmanagement. Top agrar 1, 6-8.

.

61

Chapter 4

Application of decision-tree technique to assess herd specific risk

factors for coliform mastitis in sows


Imke Traulsen, Dr. sc. agr. 1

Kerstin Reiners, Dr. sc. agr. 2

Nicole Kemper, Prof. Dr. med. vet. 3


24098 Kiel, Germany

2PIC Germany GmbH, Ratsteich 31, D-24837 Schleswig Germany



Accepted for publication in Veterinary Science Developement

62

Abstract

The aim of the study was to investigate factors associated with coliform mastitis in

sows, determined at herd level, by applying the decision-tree technique.

Coliform mastitis represents an economically important disease in sows after

farrowing that also affects the health, welfare and performance of the piglets. The

decision-tree technique, a data mining method, may be an effective tool for making

large datasets accessible and different sow herd information comparable. It is based

on the C4.5-algorithm which generates trees in a top-down recursive strategy. The

technique can be used to detect weak points in farm management.

Two datasets of two farms in Germany, consisting of sow-related parameters, were

analysed and compared by decision-tree algorithms. Data were collected over the

period of April 2007 to August 2010 from 987 sows (499 CM-positive sows and 488

CM-negative sows) and 596 sows (322 CM-positive sows and 274 CM-negative

sows), respectively.

Depending on the dataset, different graphical trees were built showing relevant

factors at the herd level which may lead to coliform mastitis. The application of birth

intervention and a higher number of piglets born alive and stillborn ones were the

main risk factors identified by the decision-tree technique to be associated with

coliform mastitis.

Herd specific risk factors for the disease were illustrated what could prove beneficial

in disease and herd management. The application of decision trees may be a

possibility of analysing critical points and decisions in management on an individual

farm basis.

Keywords

decision-tree modeling; management tool; mastitis; sows

63

Introduction

Coliform mastitis (CM) is an important infection in sows after farrowing followed by

serious economic losses due to lower productivities of the affected sows and higher

preweaning piglet mortalities (Bertschinger and Fairbrother, 2006). The diseased

animals suffer from fever and an inflammation of the mammary glands that often

leads to a decreased milk secretion 24 to 48 hours post partum. The average

prevalence in herds is about 13 %, but also a prevalence up to 60 % has been

reported (Bäckström et al., 1984; Hirsch et al., 2003; Krieter and Presuhn, 2009).

The term ‘coliform mastitis’ refers to a clinical mastitis due to coliform bacteria

(Escherichia species (spp.), Klebsiella spp., Enterobacter spp. and Citrobacter spp.)

which have been found to be associated with the disease complex in many studies

(Awad Masalmeh et al., 1990; Bertschinger and Fairbrother, 2006; Hirsch et al.,

2003; Ross et al., 1981). As a multifactorial disease, CM is influenced by the strongly

related main issues of management, feeding and hygiene as well as individual sow-

related parameters (Klopfenstein et al., 2006). It is generally assumed that optimal

herd management including the detection of weak points is a key element in reducing

the prevalence of diseases in general and CM as multifactorial infection in herds in

particular (Papadopoulos et al., 2010).

With the aid of management information technology, farmers are able to collect,

process and interpret data based at individual animal level (van Asseldonk et al.,

1999). Data mining methods are special statistical instruments which are applied to

detect relationships between attributes in datasets. The decision-tree technique, a

data mining method, has been proven as an effective tool to make large farm

datasets accessible and different sow herd information comparable (Kirchner et al.,

2004).

The aim of this study was to investigate the application of the decision-tree technique

to assess potential risk factors associated with CM-infected sows. Decision-trees

which allow deduction of association rules could support the comparison and

assessment of herd data and thereby the establishment of optimal and individual

management strategies.

64

Materials and methods

Datasets

The study was based on datasets from two rearing herds in Germany with 1,200

(Farm A) and 1,800 sows (Farm B) collected from April 2008 to August 2010 within

the scope of a microbiological study (Gerjets et al., 2010, submitted). The farms were

of high health status and tested free from porcine reproductive and respiratory

syndrome-virus, rhinitis, Actinobacillus pleuropneumoniae dysentery and enzootic

pneumonia.

The datasets comprised individual reproduction traits of the sows (Table 1) and a

respective binary record of the occurrence of CM (present or absent). All sows were

examined after farrowing and considered as mastitis cases when their rectal

temperature was above the threshold of 39.5°C 24 h post-partum (Furniss, 1987) and

the mammary glands showed definite signs of inflammation. Healthy half- or full-sib

sows from the same farrowing group served as controls. The half-sib design was

chosen due to further studies on the genetic background via genotyping (Preißler et

al., unpublished data). Manual obstetric measures after the beginning of birth were

defined as the trait ‘birth intervention’. ‘Birth induction’ was the hormonal induction of

birth after the 115. day of gestation in order to get the birth process started.

The first dataset (Farm A) consisted of a total of 987 observations – 499 observations

from CM-positive sows and 488 observations from CM-negative sows. The second

dataset (Farm B) contained 596 observations whereas 322 observations

distinguished CM-positive sows and 274 observations CM-negative sows. The mean

number of parities per sow was 4.0 for Dataset A and 3.2 for Dataset B (Table 1).

The average number of piglets born alive was 12.1 (Dataset A) and 12.3 (Dataset B)

and the average number of stillborn piglets 1.2 (A) and 1.0 (B), respectively. The

mean number of weaned piglets was 10.6 for both datasets.

65

Table 1: Means (standard deviations) and frequencies (yes/no) of reproductive traits for Farms A and B

Variable (abbreviation) Farm A (n = 987) Farm B (n = 596)

Number of parities per sow (np) 4.0 (1.9) 3.2 (1.9)

Piglets born alive per litter (pba) 12.1 (3.0) 12.3 (3.1)

Piglets born dead per litter (pbd) 1.2 (1.6) 1.0 (1.5)

Birth intervention (biv) 212/ 775 143/ 453

Birth induction (bid) 409/ 578 356/ 240

Decision tree algorithm

The C4.5-algorithm was used to generate decision trees by employing the top-down

and recursive-splitting technique (WEKA, 3-6-2, 2010). Every decision-tree consisted

of a root node and internal nodes representing the attributes, and branches that

characterized the attribute values. In this study, the reproduction parameters and the

information of birth intervention (biv) and birth induction (bid) served as attributes.

The leaves (leaf node of the decision tree) expressed the binary decision (presence

or absence of CM) and indicated the classification of either positive (CM-positive

sow) or negative (CM-negative sow) examples.

Therefore, the classification was performed by starting from the root node until

arriving at a leaf node. The descending order of the attributes within the decision-tree

and the threshold values of the branches were calculated by the algorithm with the

gain ratio criterion whereas the root of the tree represented the attribute with the

highest information gain.

In order to reduce the chance of overfitting, the C4.5-algorithm simplifies very highly

and complexly generated trees by the error-based pruning method (Quinlan, 1993).

The C4.5-algorithm is described in detail by Quinlan (1993) and Mitchell (1997).

The classification accuracy of the algorithm was tested with the stratified 10-fold

cross-validation method which analyses the number of correctly and incorrectly

classified instances (observations) (Kohavi, 1995). The whole dataset was randomly

divided into ten subsets, nine parts being dedicated to the training and one for the

test. The training set learned the algorithm and generated the tree and the test set

estimated the classification parameters. Then the algorithm ran ten times, each time

with a different training and test set, and the results were validated.

66

The classification accuracy assessment was calculated with a two-dimensional

confusion matrix consisting of the numbers of true positive (TP), false negative (FN),

true negative (TN) and false positive (FP) classified examples. Sows with CM

described the positive instances and healthy sows represented the negative

instances in this study. The classification accuracy of the C4.5-algorithm was

expressed by specific evaluation parameters (Table 2). The overall classification

accuracy described the number of correctly classified instances in total. The

proportion of correctly classified CM-positive sows in relation to all CM-positive sows

was represented by the sensitivity. In addition, the specificity was defined by correctly

classified CM-negative sows in relation to all CM-negative sows. The Kappa value

reflected the degree of agreement for classifying the sows in the CM-positive or CM-

negative classes. The error rate indicated the falsely classified CM-positive sows in

proportion to all positively classified sows.

Table 2: Evaluation parameters of the classification accuracy of the C4.5-algorithm

Evaluation parameters Formula

Classification accuracy TP+TN/(TN+FP+FN+TP) x 100

Sensitivity TP/(TP+FN) x 100

Specificity TN/(TN+FP) x 100

Kappa value (TP+TN) - [((TP+FN) x (TP+FP) + (FP+TN) x (FN+TN))/N]/

N- [((TP+FN) x (TP+FP) + (FP+TN) x (FN+TN))/N] x 100

Error Rate FP/(FP+TP) x 100

TP = true positive; TN = true negative; FP = false positive; FN = false negative, N = total number of instances.

In this study, the minimum number of instances per class varied between 20, 50 and

100, i.e. a new branch was created by the C4.5-algoithm only when it contained a

number of instances greater or equal to the adjusted values of 20, 50 and 100. The

results, calculated with the different minimum number of instances per class, were

named according to the datasets A20, A50, A100 and B20, B50, B100, respectively.

67

Results

The evaluation parameters varied between the two datasets and due to the specified

number of instances per class (Table 3). The best values were achieved for Dataset

A when the number of instances was set to the minimum of 100 instances per class

and for Dataset B when the number of instances was set to the minimum of 20

instances per class. The evaluation parameters for B20 showed a better fit compared

to A100: The classification accuracy (61.2 %) and the sensitivity (65.8 %) of B20 were

higher than for A100 (55.0 %; 58.1 %) and the error rate of B20 was 8.5 % points lower.

The Kappa value (21.7 %) of B20 reached higher values compared to A100 (10.0 %).

The specificity of B20 (44.2 %) was lower than for A100 (48.2 %).

Table 3: Evaluation parameters for Farms A (n = 987) and B (n = 596) with varied adjusted minimum

number of instances per class

Dataseta Classification

accuracy (%)

Sensitivity

(%)

Specificity

(%)

Error

rate (%)

Kappa

statistic (%)

No. of

leaves

No. of

nodes

A20 53.2 54.7 48.4 46.4 6.4 5 9

A50 54.2 55.9 47.5 45.4 8.4 4 7

A100 55.0 58.1 48.2 44.8 10.0 4 7

B20 61.2 65.8 44.2 36.3 21.7 8 15

B50 60.2 65.5 46.0 37.4 19.6 4 7

B100 56.4 64.0 52.6 41.1 11.5 3 5 a20, 50, 100 = at least 20, 50 or 100 instances per class

Graphical trees are presented for A20, A100, B20 and B100 (Figures 1, 2, 3 and 4).

The decision-trees of both datasets showed differences, although the available

attributes (‘parity number’, ‘piglets born alive’, ‘piglets born dead’, ‘birth intervention’,

birth induction) were the same for all trees

The attribute ‘birth induction’ did not appear in any of the trees showing that the other

parameters are more important for the occurrence of coliform mastitis. The attribute

‘parity number’ was not chosen in the trees of Dataset A. The trees of A20, A100 and

B20 started with the attribute ‘birth intervention’ as the root node which, therefore, was

identified as the most influencing attribute.

In Dataset A20, sows with no ‘birth intervention’ but ‘piglets born dead’ greater than

zero and ‘piglets born alive’ greater than 14 were CM-positive. In Dataset B20, sows

68

with no ‘birth intervention’, but a ‘parity number’ less than or equal to three, ‘piglets

born alive’ greater than twelve and ‘piglets born dead’ with at least one were CM-

positive. The right sub-tree demonstrated that sows with ‘birth intervention’ and

‘piglets born alive’ greater than nine were CM-positive.

The decision-trees of A100 and B100 were pruned, which made the decision steps

clearer and more generic. Therefore, attributes with a smaller information gain ratio

were dropped by the algorithm; important parameters endured.

The tree size of A100 was decreased by one leaf and two nodes in comparison to A20.

The tree of B100 had five leaves and ten nodes less than B20.

In Dataset A100, sows were CM-positive when ‘birth intervention’ was applied, with

more than one ‘piglet born dead’ or more than 14 ‘piglets born alive’. In Dataset B100,

sows with ‘piglets born alive’ greater than ten and a ‘parity number’ less than or equal

to three were CM-positive.

69

Figure 1: Decision tree showing the detected parameters and threshold values associated with CM of

dataset A20 (n = 987; minimum number of 20 instances per class); biv = birth intervention; pbd = piglets

born dead; pda = piglets born alive.


dataset A100 (n = 987; minimum number of 100 instances per class); biv = birth intervention; pbd = piglets

born dead; pda = piglets born alive.

70


dataset B20 (n = 596; minimum number of 20 instances per class); biv = birth intervention; np = parity

number; pbd = piglets born dead; pda = piglets born alive.


dataset B100 (n = 596; minimum number of 100 instances per class); biv = birth intervention; np = parity

number; pbd = piglets born dead; pda = piglets born alive.

71

Discussion

The main objective of the study was the analysis of potential risk factors associated

with sows suffering from coliform mastitis, determined on farm basis, by applying the

C4.5-algorithm of the decision-tree technique. Environmental influences were

standardised through the recording of data on these three farms with their high health

standards.

According to the microbiological and genetic study design, only clear cases of CM-

positive and selected cases of CM-negative sows were used for the analysis.

Therefore, it is not possible to make statements of the real prevalence of CM on the

farms where a large grey area of diseased sows exist.

The values of the evaluation parameters of the C4.5-algorithm were not acceptable

compared to other studies. The sensitivity and specificity were too low and the error

rate was too high. Kirchner et al. (2004) analysed culling strategies in swine breeding

data by using the decision-tree technique and reached a classification accuracy

value of about 85 %. The specificity was around 97 % and the error rate on average

15 %. Those datasets, however, consisted of 14,897 and 21,818 observations, much

more than used in this study. Using more observations for model building improves

the evaluation accuracy. With lower prevalences and therewith more skewed data, it

is easier to reach higher accuracies.

The potential risk factors identified for CM by the decision-tree induction have also

been described in other studies (Bostedt et al., 1998; Hoy, 2002; Krieter and

Presuhn, 2009). A higher number of piglets born alive was associated with a higher

risk for the sows of becoming diseased. This is in accordance with findings of Bostedt

et al. (1998), in which gilts with 1.1 piglets more than healthy sows suffered

significantly more often from feverish puerperal illness, and also showed an

increased stillbirth rate. Concerning the number of stillborn piglets, our study also

supports these results. However, other researchers did not find differences in the

number of stillborn piglets between diseased and healthy sows (Mirko and Bilkei,

2004; Van Gelder and Bilkei, 2005).

Literature about the effect of the parity number on the occurrence of mastitis is

contradictory. While Baer and Bilkei (2005) found sows of higher parity (>4) having

an increased risk of suffering mastitis, other studies have described a greater mastitis

risk for lower parity sows (1. and 2. parity) (Bostedt et al., 1998; Hoy, 2003; Krieter

and Presuhn, 2009). We also found a higher risk for primiparous sows, leading to the

72

interpretation that those sows were more prone to disease. Explanations for this

might be their not fully developed immune system (Wendt, 2000; Hoy, 2006), or that

sows suffering mastitis in their first parity might be culled. Physiological hyperthermia

is also often observed in postparturient sows, especially primiparous ones, leading to

misinterpretations (Klopfenstein et al., 2006; Gerjets et al., 2008). The investigated

factor birth intervention may be helpful in order to prevent CM in sows because it is

associated with a higher risk for mastitis and can be regarded in the management.

This fact has also been reported by Bostedt et al. (1998). Manual intervention leading

to a manipulation of the birth process might have a negative influence, especially if

accompanied by insufficient hygiene.

In our study, the decision-tree technique was shown to have the ability to illustrate

confirmed influencing factors for CM. In addition, the technique was able to weight

those factors on farm basis. Individual herd and management differences were made

clear by a different order of the attributes and different threshold values of the

branches in the trees. Decision-trees, therefore, may allow exposure of individual

weak points in the management of and comparisons between farms. In the context of

multifactorial diseases, the utilisation of such a technique is shown feasible when

certain conditions are fulfilled. For practical use, graphical trees should be smaller

with clearly arranged decision steps to simplify interpretations for farmers and

consultants. The minimum number of instances per branch has to be adjusted to the

total number of instances, i.e. a small number of instances in total requires a small

minimum number of instances per branch. The quality of the classification might be

improved by including more information about management and hygiene in the

decision-tree algorithm.

Acknowledgements


Research (BMBF) in the research programme “FUGATO – Functional Genome



73

References








73.



agalactia. Reprod. Domest. Anim. 40, 544-547.

Bertschinger, H.U., Fairbrother, J.M., 2006. Escherichia coli infections. In: Straw,

B.E., D'Allaire, S., Mengeling, W.L., Taylor, D.J. (Eds.), Diseases of swine,

Iowa State University Press, Ames, pp. 431-468.

Bostedt, H., Maier, G., Herfen, K., Hospen, R., 1998. Clinical examinations on gilts



and alagactia (MMA) in sows after farrowing. Prev. Vet. Med. 5, 133-139.







Gerjets, I., Traulsen, I., Reiners, K., Kemper, N., 2010. Comparison of virulence gene

profiles of Escherichia coli isolates from sows with coliform mastitis and

healthy sows. Submitted.

Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., Witten, I., 2009. The

WEKA data mining software: an update. SIGKDD Explorations 11,1.



Hoy, S., 2006. The impact of puerperal diseases in sows on their fertility and health

up to next farrowing. Anim. Sci. J. 82, 701-704.

74

Kirchner, K., Tolle, K.H., Krieter, J., 2004. Decision tree technique applied to pig

fanning datasets. Livest. Sci. 90, 191-200.




Kohavi, R., 1995. A study of cross-validation and bootstrap for accuracy estimation

and model selection. Proceedings of the 14th International Joint Conference

on Artificial Intelligence, Montreal, Canada. Morgan Kaufmann, San Francisco,

USA.

Krieter, J., Presuhn, U., 2009. Genetic variation for MMA treatment. Zuchtungskunde.

81, 149-154.



Scand. 54, 153-161.

Mitchell, T.M., 1997. Machine learning. McGraw Hill, New York.

Papadopoulos, G.A., Vanderhaeghe, C., Janssens, G.P.J., Dewulf, J., Maes, D.G.D.,

2010. Risk factors associated with postpartum dysgalactia syndrome in sows.

Vet. J. 184, 167-171.

Quinlan, J.R., 1993. C4.5: Programs for machine learning. Morgan Kaufmann, San

Mateo, USA.



Van Asseldonk, M.A.P.M., Huirne, R.B.M., Dijkhuizen, A.A., Beulens, A.J.M., ten

Cate, A.J.U., 1999. Information needs and information technology on dairy

farms.Comput.Electron.Agric.,22,97-107.

Van Gelder, K.N., Bilkei, G., 2005. The course of acute-phase proteins and serum

cortisol in mastitis metritis agalactia (MMA) of the sow and sow performance.

Tijdschr. Diergeneeskd. 130, 38-41.

Wendt, M., 2000. So optimieren Sie das Geburtsmanagement. Top agrar 1, 6-8.

75

General Discussion

Coliform mastitis is a multifactorial disease: the underlying causes of CM are complex

and its pathophysiology has not yet been fully elucidated. Factors contributing to the

occurrence of CM can broadly be distinguished into environmental and sow-related

ones (Figure 1). This thesis focused on the one hand on bacterial pathogens involved

in the pathogenesis of CM, especially Escherichia coli (chapter two), and on the other

hand on selected sow and birth characteristics as potential, predisposing factors

(chapters three and four).

Figure 1: The multifactorial nature of Coliform mastitis (CM)

Bacterial pathogens

The term “Coliform mastitis” refers to coliform bacteria which have been isolated from

the milk of diseased sows (Awad Masalmeh et al., 1990; Kobera, 2000). Coliform

bacteria include E. coli and other lactose-splitting pathogens (Klebsiella spp.,

Enterobacter spp., Citrobacter spp.), physiologically occurring in the intestinal tract.

Escherichia coli was the organism most often found in milk as well as in mammary

tissue of sows with mastitis (Armstrong et al., 1968; Bertschinger et al., 1977; Ross et

al., 1981).

76

Considering the age of the published studies, there is a lack of recent investigations

into the topic, particularly with regard to the enormous improvements in the

methodology of the bacteriological identification of genera, species and pathotypes

over the last few decades.

In chapter two, the bacteriological analysis of milk samples, gathered from cleaned

and disinfected glands, was performed with advanced methods including the PCR

technique. The biochemical identification system of API (bioMérieux, Craponne,

France) enabled a classification and confirmation of the analysed bacteria at

subspecies level. A corresponding software package determined probabilities and T-

values for each bacterium (apiweb, bioMérieux).

Escherichia coli was identified in over 70 % of the milk samples of CM-infected sows,

which is in accordance with findings of other studies (Ross et al., 1981; Awad

Masalmeh et al., 1990; Bertschinger, 1999; Hirsch et al., 2003). However, E. coli was

also found in similar percentages in milk samples of non-infected sows – a result

which has not yet been described in literature and is difficult to interpret. To our

knowledge, the milk of healthy sows has not been subject to detailed bacteriological

analysis up to now.

A detailed analysis of virulence-associated genes of the E. coli isolates should

provide novel insights into the role of E. coli in association with mastitis in sows.

Therefore, E. coli isolates from CM-affected – and also healthy – sows’ milk were

analysed per multiplex PCR assays for the first time.

Virulence factors, expressed by virulence genes, facilitate the survival of E. coli and

their reproduction in the respective environment. Characteristic virulence genes are

possessed by different strains of E. coli.

In this study, the analysis of the virulence gene spectrum revealed only slight

differences between E. coli isolates from CM-infected and healthy sows. Particular

virulence gene profiles or specific mammary pathogenic E. coli were not detected.

Further investigations should classify E. coli isolates by phylogeny with the aid of

multilocus sequence typing. This sensitive method provides new information and a

better understanding of pathogenic E. coli strains and their phylogenetic relationship.

Previous studies on the functional analysis of E. coli strains from animals suffering

mastitis are very limited, which is in great contrast to the analysis of other E. coli

pathotypes. Finally, the respective tools and methodologies in the laboratory have

now been available for some time (Shpigel et al., 2008).

77

The bacteriological involvement in CM remains ambiguous (chapter two):

Coliform bacteria causing inflammation of the mammary glands originate from the

sow’s faecally contaminated environment and infect the glands via the teat canal

(Elmore et al., 1978; Jones, 1979; Wegmann et al., 1986; Baer and Bilkei, 2005).

The results of this thesis support the hypothesis that sows only develop clinical signs

of CM if further adversely environmental, genetic or other influencing factors

promoting infection are present.

Other, better defined influencing factors are for instance:

Sow- and birth-related factors

Chapters three and four paid attention to sow characteristics influencing the actual

outbreak of CM. It was found that the odds of CM in swine increased with a higher

number of piglets born alive and stillborn piglets, with the application of manual birth

intervention and for gilts. Moreover, previous CM increased the chance of suffering

another CM disease. Only older investigations exist with which to compare these

results. However, the identified risk factors are physiologically consequent and

explicitly described in chapter three.

Consideration of these findings in practice could prove useful in disease and herd

management, and may be implemented in knowledge-based management

information systems in pig production. Optimal herd management is the key element

in reducing the prevalence of diseases in herds. This also includes the detection of

weak points. In chapter four, a possibility of analysing critical points and decisions in

management on an individual farm basis is suggested by the application of decision

trees. Graphical trees are very demonstrative for the farmer or consultant, and enable

a comparison between different herd data. In the context of multifactorial diseases,

the utilisation of such a technique is shown feasible when certain conditions are

fulfilled. The more observations are available, the better the quality of the decision

rules. Only reasonable parameters should be included. The minimum number of

instances per branch has to be adjusted to the total number of instances, i.e. a small

number of instances in total requires a small minimum number of instances per

branch.

78

The time during and soon after farrowing is a very sensitive period in pig production

demanding great attention by the farmers. The sows are exposed to a physiologically

extreme situation; their health is strongly influenced by their environment and

immune defence. At the current state of knowledge, it still remains unknown as to

why only some sows develop clinical signs of CM-infection. The outbreak of disease

is most likely due to the interaction of invading ubiquitous pathogens and the host’s

immune system. All known facts about this multifactorial pathogenesis are

summarised in chapter one of this thesis. Immune competence, including resistance

to diseases, is genetically determined (Mallard et al., 1992; Magnusson and Greko,

1998). Heritabilities for CM of approximately 10 % indicate the opportunity to use this

trait for selection (Lingaas and Ronningen, 1991; Berg et al., 2001). In pigs, resistant

genotypes have been identified for postweaning E. coli diarrhea and oedema disease

(Vogeli et al., 1999; Frydendahl et al., 2003; Reiner, 2008). ‘Genetic disease

resistance’ as a breeding tool is already applied in the United States, Canada,

Denmark and Switzerland. The identification of factors leading to individual

resistance is one main objective of future studies on infectious diseases.

A holistic view, considering the multifactorial nature of CM, is the only way to reduce

the diseases’ prevalence in problem herds.

A note on the study design and phenotype

Sampling and data collection took place in different rearing and nucleus herds in

Northern Germany, supervised by PIC Germany GmbH Schleswig. Definition of

diseased sows, selection of control sows and collection of milk samples were carried

out by one person (farm manager or veterinarian), who was briefed in detail and

monitored the herds routinely. The pig breeding company PIC governs an excellent

database with pedigree information and performance data of all animals housed on

the farms that was also the basis for the statistical analyses of this thesis.

The study design was chosen due to the microbiological investigation (see above)

and further studies on the genetic background of CM via genotyping (Preißler et al.,

unpublished data): healthy half- or full-sib sows from the same herd served as

controls for CM-diseased sows. The option of the statistical analysis was therefore a

case-control study performed by conditional logistic regression with parameters

potentially predisposing for the occurrence of CM.

79

The new genotyping technologies required an accurate phenotyping of the respective

animals. The diagnosis and definition of diseased sows is not that simple because

the symptoms of CM differ from sow to sow and are not always apparent. In this

thesis, sows were identified as CM-infected when their rectal temperature was above

39.5°C 24 h post-partum and the mammary glands showed defined signs of infection.

Moreover, the appearance and the performance of the piglets were evaluated with

regard to their behaviour and body condition. Measuring the rectal temperature after

farrowing is the most common practice used to diagnose CM in sows. The critical

temperature values range from 39.3°C to 40.5°C (Waldmann and Wendt, 2001) and

are in practice often difficult to interpret due to possible physiological hyperthermia in

postparturient sows. Therefore, the definition of the trait CM has to include further

parameters such as clinical mammary gland changes, decreased milk secretion,

reduced appetite and changes in piglets’ behaviour. Piglets of CM-affected sows are

normally characterised by restlessness due to starvation (Klopfenstein et al., 2006).

After exhaustion of their energy reserves, they commonly retreat to the warmest parts

of the farrowing crate and reduce their attempts to nurse.

Because of this diversity of symptoms, an unerring eye of the farmer is inevitable not

only to maintain animal health and productivity, but also to provide the phenotype

necessary for genetic investigations and breeding programs.

References





Syndrom) erkrankten Sauen verschiedener Tierbestände Österreichs.

Tierarztl. Umsch. 45, 526-535.



agalactia. Reprod. Dom. Anim. 40, 544-547.

Berg, P., Andersen, S., Henryon, M., Nielsen, J., 2001. Genetic variation for birth

assistance and MMA in sows and diarrhoea in their litters. 52nd Annual

Meeting of the European Association for Animal Production, Budapest.

80

Bertschinger, H.U., 1999. Escherichia coli infections. In: Straw, B.E., D'Allaire, S.,

Mengeling, W.L., Taylor, D.J. (Eds.), Diseases of Swine. Iowa State University

Press, Ames, pp. 431-468.

Bertschinger, H.U., Pohlenz, J., Hemlep, I., 1977. Mastitis metritis agalactia



Elmore, R.G., Martin, C.E., Berg, J.N., 1978. Absorption of Escherichia coli endotoxin

from the mammary glands and uteri of early postpartum ows and gilts.


Frydendahl, K., Jensen, T.K., Andersen, J.S., Fredholm, M., Evans, G., 2003.

Association between the porcine Escherichia coli F18 receptor genotype and

phenotype and susceptibility to colonisation and postweaning diarrhoea

caused by E-coli O138 : F18. Vet. Microbiol. 93, 39-51.




Jones, J.E.T., 1979. Acute coliform mastitis in the sow. Annales de médecine

vétérinaire 19, 97-101.




Kobera, R., 2000. Vergleichende Prüfung der klinischen Wirksamkeit von Cefquinom

und Enrofloxacin bei der Behandlung des Mastitis-Metritis-Agalaktie-Komlexes

der Sau (Comparative analysis of the efficacy of Cefquinime and

Enrofloxacine in the treatment of the MMA-complex in the sow) (doctoral

thesis). Universität Leipzig, Leipzig.

Lingaas, F., Ronningen, K., 1991. Epidemiological and genetical studies in

Norwegian pig herds. V. Estimates of heritability and phenotypic correlations

of the most common diseases in Norwegian pigs. Acta. Vet. Scand. 32, 115-

122.

Magnusson, U., Greko, C., 1998. Assessment of phagocytic capacity and opsonic

activity in blood and mammary secretion during lactation in sows. J. Vet. Med.

B 45, 353-361.

81

Mallard, B.A., Wilkie, B.N., Kennedy, B.W., Quinton, M., 1992. Use of estimated

breeding values in a selection index to breed Yorkshire pigs for high and low

immune and innate resistance factors. Anim. Biotech. 3, 257-280.

Reiner, G., 2008. Genetics and disease resistance. Dtsch. Tierarztl. Wochenschr.

115, 252-259.



Shpigel, N.Y., Elazar, S., Rosenshine, I., 2008. Mammary pathogenic Escherichia

coli. Curr. Opin. Microbiol. 11, 60-65.

Vogeli, P., Meijerink, E., Stranzinger, G., Bertschinger, H.U., 1999. Porcine

Escherichia coli diarrhoea and oedema disease: pathogenesis and

identification of genetic resistant breeding stack. Arch. Tierz. – Arch. Anim.

Breed. 42, 28-35.

Waldmann, K.-H., Wendt, M., 2001. Lehrbuch der Schweinekrankheiten (Handbook

of pig diseases). Parey Verlag, Stuttgart.

Wegmann, P., Bertschinger, H.U., H., J., 1986. A field study on the prevalence of

coliform mastitis (MMA syndrome) in Switzerland and the antimicrobial

susceptibility of the coliform bacteria from the milk. Proc Int Vet Pig Soc

Congr, Barcelona, Spain.

82

General Summary

The overall aim of this thesis is described by its title: Coliform mastitis in sows:

Analysis of potential influencing factors and bacterial pathogens with special

emphasis on Escherichia coli. This thesis is divided into four chapters.

In chapter one, current knowledge about the economically very important mastitis in

sows is summarised. Instead of body temperature as single indicator for CM

diagnosis and treatment, a combination of appropriate criteria should be applied to

achieve a proper diagnosis and to minimise the use of antibiotics. ‘Genetic disease

resistance’, as a new approach to disease reduction, offers promising potentialities

for prevention. In pathogenesis, there have been several hints of a predominant

influence of Escherichia coli since it is the pathogen most often isolated in previous

studies.

To support these findings, a study was carried out and described in chapter two.

The objective was to analyse milk samples of sows with and without CM for the

presence of E. coli and, subsequently, to investigate virulence-associated genes

because strains isolated from sows with CM have not yet been further investigated

via molecular biological methods. Escherichia coli were most often found, but the

prevalence and the virulence gene spectrum in both samples from CM-infected and

healthy sows did not differ significantly. Particular virulence gene profiles or specific

mammary pathogenic E. coli were not detected. These results support the hypothesis

that other causative factors seem to have greater influence on the pathogenesis of

porcine CM.

The emphasis of chapter three was, therefore, the investigation of potential

influencing factors, in particular sow characteristics. The odds of CM in sows

increased with a higher number of piglets born alive and stillborn piglets, with the

application of birth intervention and for gilts.

These conclusions indicate the need for an holistic approach, considering

management, especially documentation and selection of disease cases, as well as

environmental and animal factors, to deal with Coliform mastitis in sows.

83

Chapter four presents an instrument enabling the farmer to make sow herd data

visually accessible and to detect critical points in management. With the assistance

of decision trees, the same influencing parameters for the disease and their

relationships were illustrated, which were also identified by the case-control study of

chapter three. Such an approach distinguishing diseased from healthy sows and

predicting outcome could prove beneficial in disease and herd management and

support the establishment of optimal and individual strategies.

84

Zusammenfassung

Die Ziele der vorliegenden Arbeit lassen sich übergreifend durch ihren Titel

zusammenfassen: Coliforme Mastitis bei Sauen: Analyse von potentiellen

Einflussfaktoren und bakteriellen Pathogenen unter besonderer Berücksichtigung

von Escherichia coli. Die Arbeit ist in vier Kapitel unterteilt.

Im ersten Kapitel werden aktuelle Erkenntnisse hinsichtlich Coliformer Mastitis bei

Sauen zusammengefasst. Die Coliforme Mastitis (CM) stellt eine wirtschaftlich sehr

bedeutsame Erkrankung dar, die bei Sauen nach der Abferkelung auftritt. Zur

Diagnose mit anschließender Behandlung wird oftmals nur die rektale

Körpertemperatur gemessen und beurteilt. Aufgrund unterschiedlich ausgeprägter

Symptome sollte jedoch eine Kombination geeigneter Kriterien verwendet werden,

die eine genaue Diagnose ermöglichen und gleichzeitig den Einsatz von Antibiotika

in der Tierproduktion reduzieren kann. Zur Vorbeugung von CM bietet dabei auch der

züchterisch neue Ansatz einer ‚genetische Krankheitsresistenz‘ vielversprechendes

Potenzial. Aus ätiologischer Sicht existieren zahlreiche Hinweise auf Escherichia (E.)

coli als wesentlichen Infektionserreger, da dieses Bakterium in vorangegangenen

Studien hauptsächlich isoliert wurde.

Um die Rolle von E. coli genauer zu analysieren, wurde eine Untersuchung

durchgeführt, die in Kapitel zwei beschrieben ist. Zielsetzung dieser Untersuchung

war die Analyse von Milchproben gesunder und an CM erkrankter Sauen in Hinblick

auf das Vorkommen von E. coli. Weiterhin wurden die isolierten E. coli-Stämme auf

das Vorhandensein bestimmter Virulenzgene untersucht. Escherichia coli-Stämme

von Sauen mit CM sind bisher molekularbiologisch nicht näher differenziert worden.

Auch diese Studie ergab, dass E. coli das vorherrschende Bakterium in den

Milchproben erkrankter Sauen ist, jedoch mit ähnlicher Prävalenz auch in den

Milchproben gesunder Sauen gefunden wurde. Gleiches gilt für das Virulenzgen-

Spektrum, das sich nicht signifikant zwischen E. coli-Isolaten von CM-erkrankten und

gesunden Tieren unterscheidet. Spezielle Virulenzgen-Profile oder sogar Mamma-

spezifisch pathogene E. coli wurden nicht detektiert. Diese Ergebnisse stützen die

Hypothese, dass andere Faktoren einen größeren Einfluss auf die Pathogenese von

porciner CM haben.

85

Der Fokus des dritten Kapitels lag daher auf der Untersuchung von potenziellen

Einflussfaktoren auf CM, insbesondere individuellen Sauenparametern. Die Chance

für Sauen, an CM zu erkranken, erhöhte sich mit steigender Anzahl lebend und tot

geborener Ferkel, mit höherer Parität und mit einem Geburtseingriff.

Mit diesen Erkenntnissen wird die Notwendigkeit eines ganzheitlichen Ansatzes im

Umgang mit CM verdeutlicht, der Faktoren des Managements, insbesondere der

Dokumentation und der Selektion von erkrankten Sauen, sowie der Umwelt und des

einzelnen Tieres berücksichtigt.

Im vierten Kapitel wird ein Instrument für die Praxis vorgestellt, das Betriebsdaten

anschaulich und vergleichbar darstellen und Schwachstellen im Management

aufdecken kann. Mit der Hilfe von sogenannten Entscheidungsbäumen konnten die

gleichen Einflussfaktoren für CM, wie in Kapitel drei bereits identifiziert, und ihre

Zusammenhänge visualisiert werden. Eine solche Technik, die erkrankte von

gesunden Tieren unterscheidet und ein mögliches Auftreten von Krankheiten

ankündigt, ist von großem Nutzen für die Betriebsführung und unterstützt die

Etablierung von optimalen und individuellen Managementstrategien.

Mit den vorliegenden Untersuchungen wird ein wesentlicher Beitrag zur aktuellen

MMA-Forschung geliefert.

86

Danksagung

Mein besonderer Dank gilt meiner Doktormutter Prof. Dr. med. vet. Nicole Kemper für

die Überlassung des Themas, die sehr gute fachliche und menschliche

Unterstützung, das Vertrauen und die gewährten Freiräume sowie die Möglichkeit,

meine Ergebnisse auf Konferenzen und Tagungen im In- und Ausland vorzustellen.

Bei Herrn Prof. Dr. Joachim Krieter bedanke ich mich für die Übernahme des

Koreferats.

Dr. Imke Traulsen danke ich für die statistische Unterstützung, die netten Gespräche

und die tolle Woche zusammen mit Julia Brosig auf Kreta zur EAAP 2010. Ein

Dankeschön geht auch an Dr. Kerstin Reiners für ihre Hilfsbereitschaft und die gute

Zusammenarbeit mit PIC Germany GmbH.

Für die große Hilfe und Unterstützung im Labor danke ich Evelyn Lass, Jens

Wolfmüller, Andrea Menrath und Doris Diebel, ohne die die Bearbeitung eines

solchen Probenumfangs nicht zu bewältigen gewesen wäre.

Darüber hinaus bedanke ich mich bei allen Mitarbeitern für die schöne Zeit am

Institut und das gute Arbeitsklima. Ein großes Dankeschön vor allem an die

befreundeten Kollegen der Brunch-, Grill- und Kegelrunde, insbesondere Jan Körte,

Stefanie Hotes und Andreas Stukenborg, für die gemeinsamen Stunden und die

große Hilfe nicht nur im Rahmen der Doktorarbeit, sondern auch beim Projekt

„Eschenkamp 17“.

Mein ganz besonderer Dank gilt den Mädels aus dem „Hundecontainer“ Verena

Gonzalez Lopez, Marrin Arfsten, Lisa Kruse und Anna Albrecht für ihre moralische

Unterstützung, die wunderbare gemeinsame Zeit und die daraus entstandenen tollen

Freundschaften.

Abschließend möchte ich mich bei meiner Familie und Eckhard für ihren starken

Rückhalt und ihr Vertrauen bedanken.

87

Lebenslauf

Persönliche Informationen

� Name: Imke Gerjets

� Alter: 29 Jahre (05.10.1981)

� Geburtsort: Wittmund

� Eltern: Dietmar und Alste Gerjets

� Familienstand: fest gebunden

� Nationalität: deutsch

Schulische Ausbildung 08/88 – 06/92 Grundschule Stedesdorf

08/92 – 06/94 Orientierungsstufe Esens

08/94 – 06/01 Niedersächsisches Internatsgymnasium Esens

Juni 2001 Schulabschluss: Abitur

Studium

Praktika

10/02 – 09/05 Bachelor-Studiengang Agrarökologie an der

Universität Rostock

10/05 – 2007 Master-Studiengang Agrarökologie an der

Universität Rostock/ Studienschwerpunkt: Tierproduktion

10/06 – 02/07 Zweithörerschaft im Master-Studiengang Agrarwissenschaften

an der Universität Kiel

10/01 – 05/02 Landwirtschaftliches Praktikum an der Côte d’Ivoire

02/03 – 04/03 NABU Regionalbüro Ostfriesland

07/03 – 09/03 Milchviehbetrieb Behrens in Stedesdorf

07/04 – 09/04 Demeter-Hofgemeinschaft Grummersort

06/07 – 07/07 Laborpraktikum am Zentralen Institut des Sanitätsdienstes

der Bundeswehr Kiel (LA II Veterinärmedizin)

Berufliche Tätigkeit seit 09/07 Wissenschaftlicher Mitarbeiter am Institut für Tierzucht und

Tierhaltung der Christian-Albrechts-Universität zu Kiel bei

Prof. Dr. med. vet. N. Kemper

Coliform mastitis in sows: Analysis of potential ... · pig production, and has been reported...

Documents

Transcript of Coliform mastitis in sows: Analysis of potential ... · pig production, and has been reported...