This study can be divided into two phases. The first phase comprised experiments number 1-8 and was based on the examination of 3-week-old turkeys (Table 1). The following study variables were tested: Eimeria meleagrimitis inoculation (yes/no), Eimeria meleagrimitis dose (250, 1 600, 5 000, 10 000, 15 000, 20 000 oocysts per turkey), Clostridium perfringens inoculation (yes/no) and number of hours elapsed between Eimeria meleagrimitis inoculation and examination of the turkeys (varied from 97 to 171 hours). Three combinations of Eimeria meleagrimitis and Clostridium perfringens challenge were tested during phase 1: (a) turkeys with neither Eimeria meleagrimitis nor Clostridium perfringens challenge, (b) turkeys inoculated with Eimeria meleagrimitis alone and (c) turkeys inoculated with Eimeria meleagrimitis and subsequently with Clostridium perfringens.
The second phase comprised experiment number 9, with testing of the following four binary study variables: Clostridium perfringens inoculation (yes/no), Eimeria meleagrimitis dose (10 000 or 20 000 oocysts per turkey), turkey age (three or five weeks at necropsy) and number of hours elapsed between Eimeria meleagrimitis inoculation and examination (125 or 145 hours).
Sample size estimation was based on the assumption that a frequency of challenged individuals would show severe macroscopic intestinal lesions (primary experimental outcome) that were absent in unchallenged individuals. Frequency of lesions was likely to be associated with time of examination after challenge. Because optimal time point for examination was unknown, we examined birds on two to four occasions per experiment. The sampsi procedure of Stata 13 was used to explore sample sizes based on a significance level 0.05 and a power of 0.90. These calculations indicated sample sizes varying from 5 to 47 depending on expected lesion frequency (90-20 %) in the positive control group and whether the test was one-sided or two-sided. Although higher statistical power would have been desirable, we decided to design the study with sample sizes per time point and experiment that varied between five and eight in most treatment groups (Table 1), corresponding to 90-75 % occurrence of severe intestinal lesions in the positive control group. The statistical power was increased by merging treatments groups within experiments (Tables 4 and 5, and Table 2: Turkey age). Furthermore, because we conducted a total of nine experiments, we could increase sample size by merging observations from identical treatments in different experiments (Tables 2 and 3).
Animals, housing and feed
Day-old male turkeys (B.U.T. 10 in experiments 1-8, B.U.T. Premium in experiment 9) were supplied by a commercial hatchery (Baastad Kalkun, 1866 Båstad, Norway). The birds arrived in one or two transportation boxes, and were selected at random for allocation to experimental groups. In each experiment, all birds originated from the same parent flock. One day before challenge, bird size of all treatment groups was compared visually, and birds were moved in order to make the average bird size of each treatment group as equal as possible. All turkeys were kept on floor covered by wood shavings (new at the beginning of each experiment) during the entire experimental period, and were housed either in up to 21 cages (0.300 m2) or up to 10 pens (1.135 m2). Cages were used in experiments 1-4 and pens were used in experiments 5-9. Maximum estimated total live weight per m2 during each experiment was 10.0 kg and 10.9 kg for cages and pens, respectively. Total live weight estimates were based on Commercial Performance Objectives of B.U.T. 10 and B.U.T. Premium (http://www.aviagenturkeys.com/en-gb/documents) as well as recordings of actual live weights. The birds had free access to drinking water and feed, continuous light during the first 24 hours after housing and an eight-hours darkness/16-hours light cycle during the rest of each experiment.
All turkeys in this study were euthanized, either because of disease/disorders, retarded growth or (in most cases) sampling for data collection. Birds were euthanized by blunt force trauma to the head, to render unconsciousness, followed by cervical dislocation.
Room temperature was monitored daily and adjusted with bird age according to guidelines from the hatchery. No anaesthetics or surgical interventions requiring anaesthetics were used in this study. All experiments, including the method of euthanasia, were approved by the Norwegian Food Safety Authority (FOTS applications ID 5373, 5394 and 10491).
Experimental feeds were used in experiments 1-7. In experiments 1-6 a starter feed was used until day 15-17, followed by a grower feed until the end of the experiment. In experiment 7 the grower feed was used from day 0 throughout the experiment. Major ingredients of the experimental starter feed were maize, dehulled oats, extracted soybean meal and fishmeal. Major ingredients of the experimental grower feed were wheat, barley, fishmeal, extracted soybean meal and animal fats. In experiments 8 and 9 all birds were offered a single type of commercial feed for meat type poultry. Major ingredients were wheat, oats, extracted soybean meal, maize grits, maize gluten, fishmeal, animal fats and vegetable fats. None of the used feeds were supplemented with any antibiotics or anticoccidial compounds.
Preparation of inoculum and inoculation
Two strains of Eimeria meleagrimitis were used. Only one strain was used in each experiment. Strain pM3 was used in experiments 1-8 and strain USMN08-01-Line 5 was used in experiment 9. Strain pM3 was obtained from a turkey farm in Brittany, France and purified from one single oocyst in 1999. The strain has been maintained at the Ploufragan-Plouzané-Niort Laboratory of Anses (French agency for food, environmental and occupational health and safety) by regular passages in turkeys free from coccidia since it was obtained. Strain USMN08-01-5 was obtained from a turkey farm in Minnesota, USA in 2008, isolated and propagated at the University of Arkansas in Fayetteville, USA, and further characterized at the University of Guelph in Ontario, Canada . Oocysts were kept in potassium dichromate suspension at 4◦ C until the inoculum was prepared within 24 hours of use. In experiments 1-7 the oocyst suspension was centrifuged at 1300xg for 10 minutes, the supernatant with potassium dichromate was removed and the oocysts were resuspended in PBS. Oocyst concentration was then estimated based on counts in a Fuchs Rosenthal chamber, and adjusted by adding PBS if necessary. Suspensions were thoroughly homogenized immediately prior to every step of sampling and dilution. Oocysts with morphologic signs suggesting lack of infectivity/viability were not included in the counts. At least six samples from each dilution were counted. Inoculum in experiments 8 and 9 were prepared in the same way as in experiments 1-7 with the following exceptions: the original suspension was not centrifuged and potassium dichromate was not removed.
A total of five Clostridium perfringens isolates (001, 004, 013, 036 and 58702) were used. All of these isolates were recovered from severe intestinal lesions in commercial Norwegian turkeys diagnosed with necrotic enteritis. Isolate 01 was used in one experiment, isolate 04 in all experiments, isolate 13 in five experiments, isolate 36 in two experiments and isolate 58702 in one experiment. In experiments 1-8 each Clostridium perfringens -challenged bird was inoculated with only one Clostridium perfringens isolate. In experiment 9 each challenged bird was inoculated with a mix of isolates 04 and 13.
Clostridium perfringens was inoculated into 200 ml pre-reduced and pre-warmed Brain Heart Infusion broth (BHI) and incubated anaerobically at 37o C overnight (15hours) when it was diluted 1 in 2 with fresh BHI. This inoculum was introduced to the birds within 2 hours of preparation. Clostridium perfringens challenge was done on day three to six after Eimeria meleagrimitis inoculation, in most cases (see Table 5) on days 3, 4 and 5 after Eimeria meleagrimitis inoculation.
Eimeria meleagrimitis and Clostridium perfringens inoculates were administered into the crop of each bird, using a flexible plastic tube fitted onto a syringe. Low-dose Eimeria inoculation was conducted prior to high-dose inoculation. Hygienic measures were taken to avoid cross-contamination between treatment groups. Control birds were left un-inoculated.
Sampling, data recording and laboratory analyses
A total of 810 turkeys were moved from the experimental facility and euthanized (see above for details), examined and sampled during varying time periods following Eimeria meleagrimitis and Clostridium perfringens inoculation. The turkeys were put down and examined at 21 to 36 days of age. The small intestine between the gizzard and Meckel’s diverticulum of all birds was opened and inspected closely for mucosal lesions. If severe lesions were detected in the posterior part of the opened intestinal segment, the unopened part of the small intestine was opened and inspected for lesions. Lesions that were considered severe or suggestive of necrotic enteritis were recorded. At least one digital photograph of the opened intestine of each turkey was taken and used for the development of a scoring system and subsequent final scoring of each bird. The turkeys were examined for intestinal lesions on days 4-7 (about 97-171 hours) after Eimeria meleagrimitis inoculation, or at corresponding age in non-inoculated turkeys. A majority of the turkeys were examined on day 5 or 6 after Eimeria meleagrimitis inoculation, in most cases corresponding to the last day of Clostridium perfringens inoculation and the day after (Table 1).
Intestinal contents (posterior small intestine from Meckel’s diverticulum to the caeco-intestinal junction, caeca and colorectum) were pooled from birds that were sampled on same occasion and had identical combination of Eimeria meleagrimitis and Clostridium perfringens challenge (including Clostridium perfringens isolate). Collectively these 152 pooled samples originated from 97.2 % (787/810) of all turkeys that were examined for intestinal lesions. The mean number of individuals contributing to each pooled sample was 5.2. OPG counts were calculated based on routine laboratory procedures using a modified McMaster flotation technique and Whitlock Universal McMaster counting chambers.
Scoring system development and scoring
Development of the macroscopic lesion scoring system was based on written records of necropsy findings and inspection of 2312 photographic images of opened intestines from the examined birds. A four tier scoring system was devised, graded by lesion severity. A score 0 was given to turkeys with a small intestine considered macroscopically normal. A score 1 was given to turkeys with the mildest lesions, and a score 3 was given to turkeys with the lesions considered most severe. Important aspects used to form the four lesion categories include characteristics of the intestinal wall, mucosa, mucosal villi, intestinal contents and materials on the mucosal surface. Scores 1 and 3 were defined by one characteristic lesion type that was not present in birds with lower scores (but could be present in birds with higher scores). Scoring of each turkey was conducted after the scoring system had been established.
Relationships between study variables and frequency of severe intestinal lesions
The experimental unit was individual turkeys. The primary experimental outcome was recorded as presence or absence of score 3 lesions (a binary variable). Outcome estimates were reported as percentage of examined turkeys with lesion score 3.
The role of six independent variables in potentially influencing the primary outcome variable was examined: Challenge (categorical, 3 combinations of Eimeria meleagrimitis/Clostridium perfringens inoculation), Eimeria meleagrimitis dose (categorical, 2 or 3 levels), Clostridium perfringens inoculation (binary), Hours elapsed between Eimeria meleagrimitis inoculation and examination (also designated ‘Hours after Eimeria meleagrimitis inoculation’) (categorical, 2 to 4 levels), turkey Age (binary) and Experiment (categorical).
The effect of each of these study variables was analysed under best possible control for the potentially influencing effect of the other independent variables (control variables) on the frequency of severe lesions. Such control was achieved by use of subsets of available data, where all control variables were kept constant and similar for each level of the study variable. As an example, the potential impact of Challenge was controlled for by including only turkeys from the same subsets of Eimeria meleagrimitis dose, Hours after Eimeria meleagrimitis and Experiments (Table 2).
Two-sided Fisher’s exact test or Pearson chi-square test (tabulate procedure, Stata/MP 14.2) were used to compare two groups, while Dunn’s test with Bonferroni adjustment (dunntest procedure, Stata/MP 14.2) was used for multiple comparisons. P-values below 0.05 were considered statistically significant.
Relationships between study variables and oocyst excretion
The experimental unit was pooled samples of intestinal contents from turkey groups defined during the analyses of relationships between study variables and severe intestinal lesions (see above, and Tables 2-4). Pooled samples were used to analyse the impact of independent variables on OPG counts. OPG was a secondary (continuous) outcome variable. Oocyst counts were log10-transformed. Outcome estimates were reported as median log10 OPG counts.
The distribution of OPG counts within each turkey group was in most cases non-normal (Shapiro-Wilk normality test, swilk procedure in Stata 14.2). OPG data were therefore analysed using non-parametric statistical methods. Kruskal-Wallis rank test (kwallis procedure, Stata/MP 14.2) was used to compare two groups, while Dunn’s test with Bonferroni adjustment (dunntest procedure, Stata/MP 14.2) was used for multiple comparisons. P-values below 0.05 were considered statistically significant.
Comparison of statistical power of outcome variables
Data from experiment 9 were used. The unit of analysis was individual turkeys. The study variable in this comparison was based on four combinations (designated groups in Table 5) of turkey age and Clostridium perfringens inoculation (0 = 3-week-old turkeys without Clostridium perfringens inoculation, 1 = 3-week-old turkey with Clostridium perfringens inoculation, 2 = 5-week-old turkeys without Clostridium perfringens inoculation, and 3 = 5-week-old turkeys with Clostridium perfringens inoculation). Three types of outcome variable were compared; frequency of birds with score 3 lesions (severe lesions) per turkey group, frequency of birds with intestinal pseudomembranes/ulcers-depressions per turkey group and median intestinal scores (0-3 as defined in the developed scoring system) per turkey group. Intestines were scored with pseudomembranes if apparently adherent mucoid and semisolid or dry material was present (e.g. as in Figure 1F). Ulcers/depressions were scored if clearly demarcated mucosal depressions with (ulcers) or without (depressions, see Figure 1H) haemorrhage were found.
Median rather than mean intestinal scores and non-parametric testing was used because intestinal scores were not normally distributed (Shapiro-Wilk normality test, swilk procedure in Stata/MP 14.2). Dunn’s test with Bonferroni adjustment (dunntest procedure, Stata/MP 14.2) for multiple comparisons was used for analysis of impact of the study variable on all three outcome variables. P-values below 0.05 were considered statistically significant.