Colibacillosis in caged layer hens: characteristics of the disease and the aetiological agent

Abstract

In Europe, outbreaks of acute mortality in layer flocks due to colisepticaemia have frequently been observed since the mid-1990s. The aims of this study were to describe the disease, to identify the serotypes of the avian pathogenic Escherichia coli (APEC) present in these outbreaks, and to detect the presence of F11 fimbriae and flagella in the isolates. For this purpose, 20 flocks with APEC-associated increased mortality and 20 control flocks matched for age were examined. Weekly mortality rates in the colibacillosis-affected flocks reached 1.71%, versus 0.30% in the control flocks. The maximum cumulative mortality over an entire colibacillosis outbreak reached 9.19%. The disease was often flock and hen house associated, with recurrent outbreaks within one round and in successive rounds in the same house. Disease was usually acute without clinical symptoms. Peritonitis with yolk material deposited in the peritoneal cavity and polyserositis were the main lesions at necropsy. O78 strains were isolated in 15 of the 20 colibacillosis flocks, and in only one of the control flocks. The majority of strains from the control flocks could not be serotyped by the 28 O-antisera used. In general, F11 fimbriae and flagella were present in the majority of the strains. F11 fimbriae were significantly more often found in O78 isolates than in the other serotypes, and are thus more often present in isolates from colibacillosis flocks. Strains positive for F11, and for F11 and flagella, were more frequently present in heart and liver of the colibacillosis-affected flocks.

Résumé

Colibacillose chez les pondeuses en cage: caractéristiques de la maladie et de l'agent étiologique

En Europe, des cas aigus avec de la mortalité chez des troupeaux de pondeuses dus à une colisepticémie ont souvent été observés depuis la mi-quatre-vingt dix. Les buts de cette étude ont été de décrire la maladie, d'identifier les sérotypes des Escherichia coli aviaires pathogènes (APEC) et de détecter la présence des fimbriae F11 ainsi que des flagelles des souches isolées. Pour cela, il a été examiné vingt troupeaux ayant présenté une augmentation de la mortalité associée aux APEC et vingt troupeaux témoins du même âge. Les taux de mortalité hebdomadaire dans les troupeaux affectés de colibacillose ont atteint 1.71% contre 0.30% chez les témoins. La mortalité cumulative maximale couvrant l’épisode de colibacillose a été 9.19%. La maladie a souvent été associée au troupeau et au bâtiment avec des cas récurrents au sein d'une bande et dans les bandes successives dans le même bâtiment. La maladie était généralement aiguë sans symptôme. A l'autopsie, les principales lésions ont été une péritonite avec un dépôt de jaune d’œuf dans la cavité péritonéale et une polysérosite. Des souches 078 ont été isolées dans 15 des 20 troupeaux atteints de colibacillose et dans seulement un des troupeaux témoins. La majorité des souches isolées des troupeaux témoins n'ont pas pu être sérotypées avec les 28 antisérums 0 utilisés. En général, les fimbriae F11 et des flagelles ont été observés pour la majorité des souches. Les fimbriae F11 ont été trouvés significativement plus souvent chez les isolats 078 que chez les autres sérotypes et ainsi ont été plus souvent présents chez les isolats des troupeaux affectés de colibacillose. Les souches positives pour F11, et pour F11 et les flagelles ont été plus fréquemment isolées du cœur et du foie des animaux des troupeaux affectés de colibacillose.

Zusammenfassung

Kolibazillose in Käfiglegehennen: Charakteristika der Erkrankung und des ätiologischen Agens

Seit Mitte der neunziger Jahre wird in Europa das Auftreten von plötzlichen Todesfällen in Legehennenherden aufgrund von Koliseptikämie häufiger beobachtet. Ziel dieser Studie war es, das Krankheitsbild zu beschreiben, die Serotypen der aviären pathogenen Escherichia coli-Stämme (APEC) in diesen Ausbrüchen zu bestimmen und das Vorkommen von F11-Fimbrien und Flagellen in den Isolaten nachzuweisen. Aus diesem Grund wurden zwanzig Herden mit APEC-assozierter erhöhter Mortalität und zwanzig Kontrollherden im entsprechendem Alter untersucht. Die wöchentlichen Mortalitätsraten in den von Kolibazillose betroffenen Herden erreichte 1.71% im Gegensatz zu 0.30% in den Kontrollherden. Die maximale kumulative Mortalität über die Gesamtdauer eines Kolibazillose-Ausbruchs betrug 9.19%. Die Krankheit war oft an eine Herde oder an ein Stallgebäude gebunden mit wiederkehrenden Ausbrüchen innerhalb eines Durchgangs und in nachfolgenden Durchgängen im selben Stall. Bei der Sektion waren Peritonitis verbunden mit Dottermassen in der Peritonealhöhle und Polyserositis die hauptsächlichen Veränderungen. O78-Stämme wurden in 15 der 20 Kolibazillose-Herden und nur in einem der Kontrollherden isoliert. Die Mehrzahl der Stämme aus den Kontrollherden konnten nicht mit den 28 verwendeten O-Antiseren serotypisiert werden. Im Allgemeinen waren F11-Fimbien und Flagellen bei der Mehrzahl der Stämme vorhanden. F11-Fimbrien wurden jedoch signifikant öfter bei den O78-Isolaten nachgewiesen als bei den anderen Serotypen, d.h. sie waren bei den Isolaten aus den Kolibazillose-Herden häufiger zu finden. Stämme mit F11 sowie mit F11 und Flagellen waren häufiger in Herz und Leber von Tieren aus den Kolibazillose betroffenen Herden vorhanden.

Resumen

Colibacilosis en gallinas de puesta en batería: características de la enfermedad y del agente etiológico

En Europa, se han venido observando brotes de mortalidad súbita debido a colisepticemia en lotes de ponedoras desde mediados de los años noventa. Los objetivos de este estudio fueron: describir la enfermedad, identificar los serotipos de Escherichia coli aviar patogena (APEC) presentes en estos brotes, y detectar la presencia de fimbrias F11 y flagelos en estos aislados. Para este propósito, se examinaron veinte lotes con un incremento de mortalidad asociada a APEC y veinte lotes control de la misma edad. El porcentaje de mortalidad semanal en los lotes afectados de colibacilosis llego al 1.71%, frente a 0.30% en los lotes control. La mortalidad acumulada máxima durante todo el brote de colibacilosis llego al 9.19%. La enfermedad se asoció frecuentemente al lote y a la granja, con brotes recurrentes en una misma puesta y en puestas sucesivas en la misma granja. La enfermedad se presentaba normalmente de forma aguda sin sintomatología clínica. Peritonitis y material de la yema del huevo presente en la cavidad peritoneal y poliserositis fueron las lesiones más importantes a la necropsia. Las cepas O78 fueron aisladas en 15 de los 20 lotes con colibacilosis, y en únicamente uno de los lotes control. La mayoría de cepas de los lotes control no pudieron ser serotipadas con el antisuero frente a 28O utilizado. En general, las fimbrias F11 y los flagelos estabn presentes en la mayoría de cepas. Las fimbrias F11 se encontraron en un número significativamente más elevado en los aislados O78 que en otros serotipos, y fueron pues más frecuentes en los aislados provenientes de los lotes con colibacilosis. Las cepas positivas a F11, y a F11 y flagelos se encontraron con más frecuencia en corazón e hígado de los lotes afectados con colibacilosis.

Introduction

Colibacillosis, caused by avian pathogenic Escherichia coli (APEC) is one of the main causes of economic losses in the poultry industry worldwide (Morris, 1989; Dho-Moulin, 1993; Yogaratnam, 1995; Elfadil et al., 1996; Barnes & Gross, 1997). Losses occur at all ages, and are due to a decrease in hatching rate, mortality, lowered production, carcass condemnation at processing and treatment costs (Goren, 1991; Gross, 1991; Barnes & Gross, 1997). Lesions associated with colibacillosis mainly consist of airsacculitis, peritonitis, polyserositis and septicaemia. Colibacillosis is usually considered a secondary disease, following a primary infection with respiratory pathogens and/or unfavourable environmental conditions (Barnes & Gross, 1997). However, it can also be a primary cause of disease (Cheville & Arp, 1978; Dhillon & Jack, 1996; Zanella et al., 2000).

Most studies on colibacillosis refer to broilers, even though laying hens can also be severely affected (Dhillon & Jack, 1996; Zanella et al., 2000). In Europe, outbreaks of acute mortality in layer flocks due to E. coli septicaemia have frequently been observed since the mid-1990s. Nevertheless, scientific reports on colibacillosis in this type of birds are scarce. The characteristics of the disease and its etiological agent have not yet been well defined. Consequently, it was the aim of this study to describe the disease related to colibacillosis in layer hens, based on field observations, and to identify the serotypes of E. coli involved.

F11 fimbriae are a type of P fimbriae that protect E. coli against phagocytosis (Pourbakhsh et al., 1997), and flagella make the strains motile. Since information on their prevalence in E. coli isolates from layers might give some perspectives towards the use of a vaccine based on these two antigens, it was an additional aim of this study to determine their prevalence in E. coli strains of layer hens.

Materials and methods

Flocks and inclusion criteria

Forty poultry flocks were included in this study. Selected flocks were Belgian commercial caged layers, producing eggs for consumption. Twenty flocks were suffering from colibacillosis while the other 20 flocks were considered controls. A flock suffering from colibacillosis was identified on the basis of: (1) a ‘farmer-reported’ increase in mortality rate, as compared with the normal baseline mortality in the flock; (2) necropsy lesions of five hens, compatible with colibacillosis; and (3) E. coli isolated from heart and/or liver in pure or abundant cultures. As soon as a colibacillosis flock was identified, a control flock with hens of approximately the same age was sought. A control flock was defined as a flock in the house of which no increased mortality as compared with the standards of the breeding organization had been observed during the current and the previous round(s).

Farmers were asked to fill out a questionnaire on the clinical symptoms seen in the colibacillosis-affected flocks, and, if any, in the control flocks. Egg production — that is, percentage lay and egg quality (% cracked and soiled eggs exceeding 1.5% or not) — and weekly mortality were monitored for at least 5 weeks prior to and 5 weeks following the sampling.

The 40 flocks were divided over 25 different farms, all in the northern part of Belgium (Table 1). On 14 farms only one flock was sampled. On eight farms two flocks were sampled, and on two farms and one farm three and four flocks were sampled, respectively. Four farms yielded both a colibacillosis and a control flock. Ages of the control flocks varied between 21 and 74 weeks, while the colibacillosis-affected flocks ranged from 21 to 87 weeks (Table 2). All farms with more than one flock were multi-age farms.

Necropsy

During a single visit per flock within 2 weeks after the outbreak for the colibacillosis-affected flocks, and at a comparable age for the control flocks, diseased hens and hens that had recently died were collected for necropsy. Routinely, five birds from all flocks suffering from colibacillosis and all available diseased or dead hens from the control flocks were examined. Gross lesions were noted and samples were collected for bacteriology.

E. coli isolation

One heart, one liver and one ovarium or oviduct were routinely sampled for bacteriology, originating from one to three animals, by preference from an animal showing macroscopical lesions in the concerned organs. The trachea was by preference taken from an animal showing lesions in one of the other organs. Depending on the localization of the lesions, the ovary or the proximal part of the oviduct was used for E. coli isolation. If both lesions were present, the ovary was preferred for sampling. If no organs with macroscopical lesions were available, as could be the case in hens from control flocks or hens from colibacillosis flocks that had acutely died, bacteriology was performed on macroscopically normal organs. The caecal contents were collected from all necropsied birds and pooled for analysis.

Without further preparatory steps, samples were inoculated on Columbia sheep blood agar (Becton & Dickinson, Erembodegem, Belgium), and incubated aerobically at 37°C. E. coli colonies were identified by their indol production using pepton water and Kovàcs Indol reagent (Merck, Darmstadt, Germany), their ability to ferment glucose with gas production, using Triple Sugar Iron agar (Oxoid, Hampshire, UK), and their inability to use citrate as a carbon source, using Simmons citrate agar (Oxoid).

Per flock, one E. coli colony from the heart, liver, trachea, and ovary or oviduct was kept for further analysis, except for the pooled caeca, from which three colonies were selected. Thus a minimum of three and a maximum of seven isolates were obtained per flock. They were stored in Luria Bertani broth containing 50% glycerol at −80°C until further analysis.

O-serotyping

Twenty-six different antisera (O1, O2, O5, O6, O8, O9, O11, O12, O14, O15, O17, O18, O20, O35, O36, O45, O53, O78, O81, O83, O88, O102, O103, O115, O116, O132), purchased at the E. coli reference centre of the University of Santiago de Compostela (Spain), were used to serotype the E. coli strains, which was done using the microtitre agglutination test, as described by Guinée et al. (1972). Supplementarily, an O109 antiserum, produced in a rabbit by our laboratory (Veterinary and Agrochemical Research Centre, Brussels, Belgium), and an O157 antiserum (Statens Serum Institute, Copenhagen, Denmark) were used on those strains that were non-typable by the aforementioned 26 antisera. Here the slide agglutination technique was used. If a strain could not be serotyped by the 28 available sera, it was classified as non-typable (NT).

F11 fimbriae and flagella detection

F11 fimbriae detection was carried out with an indirect enzyme-linked immunosorbent assay as described by van den Bosch et al. (1993). The presence of flagella was detected by assessing the motility of the isolates in MIO Medium (Becton Dickinson, Erembodegem, Belgium).

Statistical analysis

Logistic regression (Glimmix macro; SAS, version 8.02) was performed on the data, to examine possible associations between the type of flock (i.e. control or colibacillosis-affected flock), weekly mortality and egg production data. Weekly mortality was expressed as the proportion of dead hens per week by the number of hens originally present in the flock at the age of 17 weeks. Percentage lay was the daily number of eggs produced divided by the number of hens present in the flock at that time. Egg quality was described as a dichotomous variable: normal or increased rate of second choice eggs (threshold 1.5%). Fisher's exact test (Frequency procedure; SAS, version 8.02) was used to compare the frequency of the O78 and NT serotypes, of E. coli isolation from heart and liver, of the presence of F11 fimbriae in isolates from heart, trachea, and ovary/oviduct, and of flagella in isolates from heart, liver, trachea and ovary/oviduct between control and colibacillosis-affected flocks. Pearson's chi-square (Frequency procedure; SAS, version 8.02) was used for the data concerning E. coli isolation from trachea and ovary/oviduct, for F11 presence in isolates from the livers and pooled caeca, and for the presence of flagella in the pooled caeca. For the presence of F11 and flagella, the Pearson's chi-square test was used for all organs except the trachea and ovary/oviduct, for which the Fisher's exact test was used. Additionally, possible associations between the presence of F11 fimbriae and flagella in the isolates, and flock type, organ and O-serotype were examined (Glimmix macro; SAS, version 8.02).

Results

Clinical symptoms

A detailed description of the characteristics of control flocks and flocks suffering from colibacillosis is presented in Tables 1 and 2.

Table 1. Bacteriological results of the animals originating from control flocks and from colibacillosis-affected flocks

						E. coli isolated from
										Pooled caeca
Flock number	Farm	Number of flocks on farm	Breeding company	Age at sampling (weeks)	Number of E. coli isolated	Heart	Liver	Trachea	Ovarium/oviduct	1	2	3
Control flocks
1	24	1	3	21	3	–^b	–	–	–	O15	O15	NT^c
2	21	2	3	21	6	O2	NT	–	O2	O2	O2	O2
3	19	2	3	22	3	–	–	–	–	NT	NT	NT
4	20	2	3	26	3	–	–	–	–	NT	NT	NT
5	21	2	3	27	5	O20	O78	–	–	NT	NT	NT
6	18	3	3	28	5	NT	NT	–	–	NT	NT	NT
7	8	1	2	33	3	–	–	–	–	NT	NT	NT
8	19	2	3	36	4	–	–	O2	–	NT	NT	NT
9	18	3	3	36	7	NT	NT	O88	NT	O109	NT	NT
10	19	2	1	36	6	O5	O5	–	O5	O2	O5	NT
11	25	2	3	36	3	–	–	–	–	O8	O8	O8
12	7	3	2	39	6	O88	O88	–	O88	NT	NT	NT
13	11	1	2	41	5	–	NT	NT	–	NT	NT	NT
14	6^a	2	1	43	7	O83	O17	O17	O83	O83	O83	O83
15	5^a	2	2	55	4	NT	–	–	–	O5	O6	NT
16	15^a	3	2	56	5	–	O2	O18	–	NT	NT	NT
17	7	3	2	57	4	–	NT	–	–	NT	NT	NT
18	4^a	4	2	58	4	–	–	–	NT	NT	NT	NT
19	19	2	3	59	7	O2	O2	O2	O2	NT	NT	NT
20	18	3	3	74	6	O53	NT	–	NT	NT	NT	NT
Colibacillosis-affected flocks
1	23	2	3	21	7	O78	O78	O8	O78	O6	O6	O78
2	9	6	2	22	7	O78	O78	O78	O78	O78	O78	O78
3	22	1	3	22	7	O2	O2	O2	O2	O2	NT^c	NT
4	17	2	2	25	7	O78	O78	O78	O78	O78	O78	NT
5	4^a	4	2	26	7	O78	O78	O78	O78	O78	O78	O78
6	12	2	2	29	7	O78	O78	O78	O78	O78	NT	NT
7	1	3	1	33	6	O78	O78	O78	–	O78	O78	O78
8	5^a	2	2	36	7	O78	O78	O78	O78	O78	O78	O78
9	9	6	2	36	6	O78	O78	–	O78	O78	O78	O78
10	12	2	2	36	6	O78	O78	O78	–	O78	O78	O78
11	16	5	2	36	7	O78	O78	O78	O78	NT	NT	NT
12	13	2	1	38	6	O78	O78	O78	–	O78	O78	O78
13	6^a	2	1	39	6	O2	O2	O2	–	O2	O5	NT
14	10	1	1	48	6	O18	O18	O18	–	O18	O18	O18
15	2	1	1	53	6	O2	O78	–	O2	O78	NT	NT
16	9	6	2	53	5	O78	O78	–	–	O14	O78	O78
17	3	2	2	54	7	O78	O78	O78	O78	O14	O78	O78
18	16	5	2	59	7	O18	O18	O78	O18	O18	O78	NT
19	15^a	3	2	61	4	–	O83	–	–	O8	O8	O8
20	14	1	2	87	7	NT	NT	NT	NT	O6	O6	O9
^a Farms where a control and a colibacillosis-affected flock were sampled.
^b No E. coli isolated.
^c NT, non-typable with the 28 O-antisera used.

Table 2. Characteristics of the control and colibacillosis-affected flocks

	Control flocks	Colibacillosis-affected flocks
Number of flocks sampled	20	20
Number of farms sampled	13	16
Flock size	11,005 to 44,000	11,700 to 50,000
Age at sampling (weeks)	21 to 74	21 to 87
Breed
Isa Brown	10	17
Lohman Selected Leghorn	3	3
Lohman Brown	3	0
Hisex Brown	3	0
Hisex White	1	0
Maximum weekly percentage of mortality	0.07 to 0.30	0.26 to 1.71
Duration increased mortality in weeks	–	3 to 10
Clinical signs	0	11
None	20	10
Depression	0	5
Soiled cloacae	0	6
Decrease in egg production>2%	1	6
Percentage decrease in egg production	2.52	2.20 to 5.65
Increased rate of second choice eggs	0	3
Increase in second choice eggs and decrease in egg production>2%	0	1

In the colibacillosis-affected flocks, the weekly mortality, which was a selection criterion, increased with a factor of three to eight within a 1-week to 3-week period following the onset of the outbreak. Peak mortality ranged from 0.26% to 1.71% per week. This was significantly different (P<0.0001) from the peak mortality in the control flocks, ranging from 0.07% to 0.30%.

Nine of the affected flocks were sampled during the first colibacillosis episode of the round, and 11 flocks during a recurrent outbreak. In three flocks, a total of four periods of increased mortality had been observed at that time. Of the 11 flocks showing a recurrent outbreak of E. coli-caused mortality, two had been treated with antibiotics (flumequine and tetracyclin), and one with a homeopatic product. The other eight flocks had not received any treatment.

The first notifications of increased mortalities varied between 21 and 55 weeks of age in the different flocks, but in 70% of the cases the first outbreaks were seen at the onset of lay and peak production (i.e. 20 to 40 weeks of age). Increased mortality lasted for a minimum of 3 weeks. In five flocks, it was chronic, lasting for over 10 weeks. In one of these flocks, the cumulative mortality over this period reached 9.19%. For eight colibacillosis flocks, the farm managers reported a local increase in mortality in the hen house at the start of the colibacillosis outbreak, after which it spread throughout the batteries. In 10 colibacillosis flocks, the increased mortality was from the start evenly spread over all batteries. Only in two affected flocks, the mortality remained localized in certain parts of the batteries.

Symptoms were seen in 10 colibacillosis flocks (i.e. depression, and/or soiled cloacae), but only in a limited number of hens. A decrease in egg production of 2.20% to 5.65% at the time of maximum mortality and an increase in the rate of second choice eggs was seen in six and three colibacillosis flocks, respectively. In only one flock, both these symptoms were seen. No significant difference was found for the egg production data between the colibacillosis flocks at the age of maximum mortality and the control flocks at a comparable age.

No potential stressors, like malfunctions in the house or extraordinary temperatures, were reported for any of the flocks, occurring shortly before the colibacillosis outbreak in the affected flocks or shortly before the sampling in the control flocks.

Necropsy lesions

The lesions found in necropsied hens derived from the control flocks were very diverse, including abscesses, dystocia, liver lesions and bone trauma. Articular amyloidosis was present in two flocks. Polyserositis was seen in birds from two flocks, peritonitis with yolk material deposited in the peritoneal cavity in birds from five flocks.

In the flocks suffering from a colibacillosis outbreak on the other hand, the birds consistently showed lesions of polyserositis, including perihepatitis (11/20), pericarditis (12/20) and peritonitis with yolk material deposited in the peritoneal cavity (19/20). Furthermore, oophoritis was observed in 8/20 flocks and salpingitis in 6/20. Liver lesions (ruptured liver, haematoma, steatosis or fragile liver) were found in four flocks, and dystocia, femoral head problems and inactive ovarium each in three flocks. Egg concrements in the oviduct, urate deposits in the viscera, catarrhal enteritis and bone trauma were each seen in only one flock.

Characteristics of E. coli strains

Following necropsy, 224 E. coli strains were isolated. Ninety-six strains were derived from control flocks and 128 from flocks suffering from colibacillosis (Table 3). E. coli was significantly more often found in the heart (19/20, P=0.0033), liver (20/20, P=0.0033) and trachea (16/20, P=0.0036), from colibacillosis flocks than from control flocks, for which the heart, liver and trachea were positive for E. coli in 10, 12, and six of the 20 flocks, respectively. The difference between both types of flocks for isolates from ovary or oviduct (i.e. 13/20 in colibacillosis and 8/20 in control flocks) was not significant (P=0.2049).

The results of the bacteriological analyses per organ and the serotyping of isolates from the control and colibacillosis flocks are presented in Table 1.

Among the isolates from flocks experiencing outbreaks of colibacillosis, the O78 serotype was predominant, being isolated from 15/20 flocks (75%). Serotype O78 was isolated from only one of the control flocks (5%). This difference between both types of flocks was significant (P<0.0001). The majority of strains obtained from the control flocks was not typable with the set of antisera used. NT strains were isolated from 18 out of 20 control flocks (90%). NT strains were also found in 8/20 (40%) of the flocks suffering from colibacillosis, yielding a significant difference with the control flocks (P=0.0022). Serotypes O9 and O14 were only found in colibacillosis-affected flocks, and O15, O17, O20, O53 and O109 were only found in control flocks. O2, O5, O6, O8, O18, O78, O83, O88 and NT were found in control as well as in colibacillosis flocks. Insufficient data were available on the identity of the flocks’ rearing farms to allow a close epidemiological examination of the encountered serotypes. Per breeding company, six to 10 different O-serotypes were found, besides the NT strains (Table 1). Of the 15 different serotypes found in total, nine were present in at least two of the three breeding companies. The six serotypes ‘exclusive’ to a certain breeding company were detected in only one or two flocks each.

The strains isolated from the pooled caeca in 17/20 colibacillosis flocks belonged at least in part to the same serotype as the strains isolated from the other organs. This phenomenon was also seen in 10/20 of the control flocks. The serotypes O17, O20, O53 and O88 were detected only among extraintestinal isolates, and O6, O9, O14, O15 and O109 were only detected among caecal isolates.

For the control and colibacillosis-affected flocks situated on the same farm, different serotypes were detected, even though the hen houses were situated on the same premises (Table 1).

The presence of F11 was demonstrated in 65/96 (68%) E. coli isolates from control flocks and in 111/128 (87%) strains isolated from colibacillosis flocks (Table 3). Statistical analysis revealed that this prevalence was significantly higher for the colibacillosis flocks (P=0.0188) due to the association of F11 with the O78 serotype (P=0.0013). Table 3 shows that 17/19 (90%) of the heart isolates in the colibacillosis flocks are positive for the presence of F11, versus 5/10 (50%) in the control flocks, yielding a significant difference (P=0.0302). For the isolates from the pooled caeca, 53/60 (88%) were positive for F11 fimbriae in the colibacillosis flocks versus 44/60 (73%) in the control flocks. This difference was also significant (P=0.0369). For the isolates from the liver (P=0.0763), the trachea (P=0.5407), and the ovary/oviduct (P=1.0000), no significant differences were found between both types of flocks.

Table 3. Distribution of the isolates per flock type and organ for the detection of F11 and flagella.

		Number of isolates per organ of origin
	Total number of isolates	Heart	Liver	Trachea	Ovary/oviduct	Pooled caeca
Control flock isolates	96	10^a	12^b	6^c	8	60
Average number of isolates per flock	4.8	0.5	0.6	0.3	0.4	3.0
Number of isolates positive for F11 (%)	65^d (68)	5^e (50)	6 (50)	4 (67)	6 (75)	44^f (73)
Number of isolates positive for flagella (%)	84 (88)	9 (90)	11 (92)	5 (83)	7 (88)	52 (87)
Number of isolates positive for F11 and flagella (%)	53 (55)	4 (40)	5 (42)	3 (50)	5 (62)	36 (60)
Colibacillosis-affected flock isolates	128	19^a	20^b	16^c	13	60
Average number of isolates per flock	6.4	0.9	1.0	0.8	0.6	3.0
Number of isolates positive for F11 (%)	111^d (87)	17^e (90)	16 (80)	14 (88)	11 (85)	53^f (88)
Number of isolates positive for flagella (%)	100 (78)	14 (74)	17 (85)	12 (75)	9 (69)	48 (80)
Number of isolates positive for F11 and flagella (%)	87 (68)	13 (68)	15 (75)	10 (62)	7 (54)	42 (70)
^a Difference is significant in the Fisher's exact test (P=0.0033).
^b Difference is significant in the Fisher's exact test (P=0.0033).
^c Difference is significant in the Pearson's chi-square (P=0.0036).
^d Difference is significant in the logistic regression (P=0.0188).
^e Difference is significant in the Fisher's exact test (P=0.0302).
^f Difference is significant in the Pearson's chi-square (P=0.0369).

Flagella were detected in 84/96 (88%) control flock isolates, and in 100/128 (78%) colibacillosis flock isolates (Table 3). The prevalence of flagellated strains was not significantly different between both flock types (P=0.2750), nor for the different organs per flock type (P _heart=0.3880, P _liver=1.0000, P _trachea=1.0000, P _{ovary/oviduct}=0.6065, and P _{pooled caeca}=0.3272). For serotype, however, significant differences in the prevalence of flagella were found between NT (63/73, 86%), O78 (52/79, 66%) and the other serotypes (69/72, 96%). O78 had the lowest percentage of isolates with flagella, yielding a significant difference with NT isolates (P=0.0112) and with the other serotypes (P<0.0001).

Eighty-seven of the 128 colibacillosis flock isolates (68%) were positive for both F11 and flagella, versus 53/96 (55%) of the control flock isolates, which was borderline significant (P=0.0509) (Table 3). The most pronounced difference was found between the isolates from the livers, of which 15/20 (75%) were positive for both factors in the colibacillosis flocks, versus 5/12 (42%) in the control flocks, giving a borderline significance (P=0.0593). For the other organs, no significant differences were found (P _heart=0.1397, P _trachea=0.6550, P _{ovary/oviduct}=1.0000, and P _{pooled caeca}=0.2508).

Discussion

Currently, only two case reports have been published indicating that colibacillosis can cause severe losses in layer hens (Dhillon & Jack, 1996; Zanella et al., 2000). Systematic studies on the clinical and pathological characteristics of the infection in this type of birds have not yet been done. The present paper therefore aims to be the first report describing outbreaks of colibacillosis in layer hens as a well-defined disease entity.

In this study 20 flocks suffering from colibacillosis were compared with 20 healthy control flocks of the same age, to deduce the characteristics of colibacillosis in layer hen flocks. Although under field conditions it can never be excluded that other diseases are simultaneously present, it became clear that outbreaks of colibacillosis are characterized by sudden deaths in layer hens, the absence of typical signs of clinical disease, and lesions of polyserositis. The condition usually initiates during early production and often reoccurs multiple times within one production cycle. Cumulative mortality due to colibacillosis may amount to more than 9% during a single outbreak. Flocks subsequently kept in the same house can also be affected. Outbreaks of colibacillosis are not systematically accompanied by a decrease in egg production or a poor egg quality.

Serotyping of E. coli isolates obtained during this study revealed that the O78 serotype was present in 15 of the 20 flocks suffering from outbreaks of colibacillosis, but in only one control flock. In several other E. coli-associated diseases like cellulitis, septicaemia and airsacculitis, O78 is a predominant serotype (Sojka & Carnaghan, 1961; Cloud et al., 1985; Dozois et al., 1992; Gomis et al., 1997). In the latter manifestations, E. coli serotypes O2 and O1 play a role too. In the present study, the O2 serotype was isolated as well, but in similar frequencies from control and colibacillosis flocks. None of the isolates belonged to the O1 serotype.

In the control flocks, the isolated E. coli strains were often NT. In the colibacillosis flocks on the other hand, NT strains were in comparison rarely found. This is probably due to the test kit we used, which is composed of 26 O-types frequently found in diseased poultry (Blanco et al., 1997). This finding supports the view that the healthy flocks involved in this study are suited control flocks, carrying E. coli types that are generally considered non-pathogenic. The NT strains found in this study probably represent several serotypes.

The O157 serotype of E. coli was not isolated in this study. O157 verotoxigenic E. coli (VTEC) strains cause severe disease in humans. In general, poultry is not considered an important source of VTEC infection for humans. No bacteriologically confirmed cases of human O157 infections originating from a poultry source have so far been reported (Chapman et al., 1997), even though broilers have been found infected with this type of VTEC (Hafez et al., 1997), and Stavric et al. (1993) found that experimental infection of layers per os lead to intestinal colonization with O157 and other VTEC. Our findings indicate that, under the circumstances of this study, colibacillosis in layer hens did not present a risk for zoonotic disease.

Until now the pathogenesis of colibacillosis in layer hens remains unclear. The route of entry leading to septicaemia has not been established. The results of this field study demonstrate that, when E. coli is present in the trachea or ovary/oviduct, often the same serotypes are detected there as in heart and liver. Moreover, E. coli was significantly more often present in the trachea of hens from colibacillosis than from control flocks. This may point to an aerogenic route of infection. The ovary/oviduct was also more frequently positive in colibacillosis flocks, but here the difference with the control flocks was not significant. Ascending infections through the oviduct therefore might evoke colisepticaemia less frequently.

For 17/20 colibacillosis flocks the serotypes of the extraintestinal and the caecal isolates were at least in part the same. This phenomenon was also found in 10/20 control flocks. The ‘prevalence theory’ (Ørskov & Ørskov, 1985) says that serogroups of extraintestinal isolates are also frequently present in the faecal reservoir. However, several authors (Whittam & Wilson, 1988; White et al., 1990, 1993; Ngeleka et al., 1996; Blanco et al., 1997) support the ‘special pathogenicity theory’ (Ørskov & Ørskov, 1985); that is, that APEC belong to special groups of pathogenic clones. Since the O78 serotype was found in 75% of the colibacillosis flocks and only in 5% of the control flocks, our results support the latter theory. Further research will be needed to assess whether the O78 strains are highly related or not. Although their incidence and site of isolation could differ, the serotypes O2, O5, O6, O8, O18, O78, O83 and O88 were found in both colibacillosis and control flocks. This might indicate that APEC can be only partly defined by serotyping, and that within one serotype strains may have a different genotype and pathogenic potential. This has already been stated for isolates from other E. coli-related diseases in poultry (Blanco et al., 1997; Carvalho de Moura et al., 2001).

Our results indicate that specific APEC strains are responsible for the colibacillosis outbreaks. No data are available determining the origin of these strains, nor how long the flock had been infected before the increased mortality started. A close examination of the epidemiological relationship between the rearing units and the flocks they produced, would be highly relevant. Unfortunately, such an analysis was not possible because only for a part of the flocks data on the identity of their rearing farms were still available at the time of sampling.

The F11 fimbrial antigen was detected in 87% of the colibacillosis flock isolates, and in 68% of the control flock isolates. This difference was even more striking when comparing its prevalence in strains isolated from the heart only: F11-positive strains were isolated in 90% versus 50% of the heart samples obtained from hens that were derived from the colibacillosis flocks and control flocks, respectively. For the other extra-intestinal organs, there were also clear differences between the control and colibacillosis flock isolates, but they were not significant. These results may indicate that F11 is a factor contributing to the virulence of E. coli strains in layer hens. F11 fimbriae are one of the 11 types of P fimbriae currently known (Low et al., 1996). In a study by van den Bosch et al. (1993), 78% of the examined APEC strains isolated from septicaemic chickens were found to express F11 fimbriae. The pap gene, encoding P fimbriae, is more frequently found in extraintestinal than in faecal isolates (Stordeur et al., 2002), and strains expressing P fimbriae are able to resist phagocytosis (Pourbakhsh et al., 1997). The exact role of F11 in colibacillosis of layer hens needs further research.

Flagella were present in 78% of the colibacillosis flock isolates, and in 88% of the control flock isolates. No significant differences were found for flock type or organ of origin. They were, however, significantly less often present in the O78 isolates (66%) and in the NT isolates (86%) compared with the other serotypes (96%). This may indicate that the virulence of E. coli strains of layer hens cannot be discriminated on the basis of presence or absence of flagella.

The present study focused on characteristics of E. coli isolates involved in outbreaks of colibacillosis, and their possible relationship with virulence. However, an unfavourable housing climate as well as other infections might predispose the hens to E. coli septicaemia. Viral infections such as Newcastle disease, infectious bronchitis and avian pneumovirus, which may predispose chickens to E. coli infections of the respiratory tract, usually lead to a drop in egg production and a poor egg quality. None of the latter symptoms were consistently observed in the present study, possibly indicating that these infections did not play a role in the pathogenesis of colibacillosis in the majority of the flocks. Moreover, this was confirmed in a concurrent experiment (Vandekerchove et al., in press). The absence of stressors before the outbreak, as reported by all the farmers, further suggests that the bacteria act as primary pathogens.

Because of the high losses due to colibacillosis in layers, the needs for an effective vaccine are increasing. The only commercially available E. coli vaccine at the time of this study was a subunit vaccine intended for use in broiler breeders, containing F11 fimbrial and flagellar toxin antigens. In view of our results, this vaccine may also be useful in layers. Further research on this subject is needed.

In conclusion, colibacillosis causes important economic losses through high mortality. The disease is flock and hen house associated. In this study serotype O78 was most often isolated from the colibacillosis-affected flocks. Even though serotyping can offer a first assessment of the strains found after a sudden increase of mortality, variation in virulence is observed within one serotype. Both F11 fimbriae and the O78 serotype are significantly more often present in the isolates from the colibacillosis-affected flocks. However, F11 are also found in isolates from the control flocks, as was the O78 serotype. Further research will be needed to allow a clearer characterization of APEC and to define the predisposing factors favouring a colibacillosis outbreak.

Acknowledgments

The authors wish to thank the farm managers, breeding companies, feed firms, field practitioners and the staff of Flanders Animal Health Care for their valuable help in making this study possible. Many thanks to M. Van Hessche for her skilled technical support, and to P. van Empel for organizing the F11 and motility analyses.