Study I: Determination of MAP prevalence in ‘abnormal’ LN

The objective of this pilot cross-sectional study was to estimate the prevalence of MAP infection in ‘abnormal’ LN. Lymph nodes were classified subjectively as ‘abnormal’ by an accredited meat inspector based on grossly visible pathology, such as discolouration, focal or diffuse enlargement and/or containing a core of mineralized, caseous and/or necrotic material.

5.3.1.1 LN selection and testing

One hundred and twenty-nine ‘abnormal’ LN, sourced from 76 herds, were sampled from deer slaughtered in four South Island and three North Island DSP between 15 May 2007 and 23 November 2007. One-hundred and fourteen LN were sourced from 63 South Island herds in the Canterbury (28), Southland (18) and Otago (17) regions, whilst 15 LN were sourced from 13 North Island herds in the Waikato (5), Hawkes Bay (5), Bay of Plenty (2) and Manawatu (1) regions. A maximum of one LN per animal and three animals per line were sampled by the inspector. Two 15 mm MLN sections were stored in separate 70 ml sterile containers, one with 10% formalin. All tissues were collected by sterile technique using new equipment for each sample.

Fresh samples were kept chilled in an insulated container prior to courier to AgResearch Wallaceville (Upper Hutt, New Zealand) for culture. A liquid culture system, as described by

106 Whittington et al. (1999), using Bactec 12B vials (Becton Dickinson, Sparks, Maryland, USA) supplemented with 0.8 ml of egg yolk, mycobactin (Allied Laboratories, Fayette, Missouri, USA) and antibiotics (PANTA, Becton Dickinson, Sparks, Maryland, USA) was used (Whittington et al 1999). Subsequent to a positive culture result, a conclusive diagnosis of MAP was made based on the presence of a slow-growing acid-fast organism and mycobactin dependence of the isolate (de Lisle et al 2003).

Fixed tissue samples were dehydrated, cleared and embedded in paraffin. Sections were stained with haematoxylin and eosin (H & E) and Ziehl-Neelsen (ZN). All sections were examined by one histopathologist (RGC) without knowledge of the age, sex or herd of origin of the deer being examined.

5.3.1.2 Data collection and management

Herd, DSP, age, sex, and LN location and colour (normal, red, brown, green, yellow, white, other) were recorded at the time of sampling. Because of between-DSP variation in the precision of recording, age was aggregated into young and adult and combined with sex into categories of young hinds, young stags, adult hinds and adult stags. With the exception of herd location and DSP, data were recorded at the carcass-level.

5.3.1.3 Statistical analysis

The prevalence of MAP within ‘abnormal’ LN, nationally and by Island (North and South), was calculated by dividing the number of culture-positive LN by the total number of LN sampled, with exact confidence intervals derived from the binomial distribution (Dohoo et al 2003b). A Fisher’s exact test was used to assess the significance of herd location (North or South Island). This test ignored multiple sampling per herd as few LN were sampled per herd and, thus, only a low effect was expected on the estimated variance.

107 5.3.2 Study II: Determination of an ‘abnormal’ MLN circumference cut point and factors associated with increased MLN circumference

The objective of this study was to define the MLN circumference cut point at which the specificity of MAP detection was >95%. Associations between MLN circumference and animal- level factors and histopathological features were also evaluated.

5.3.2.1 Selection of herds and MLN

Mesenteric LN samples were collected from 412 deer slaughtered from 79 herds in three South Island (n = 49 herds) and two North Island DSP (n = 30 herds) between October 2007 and January 2009. Any herd from which five or more deer were slaughtered on any sampling day was included. Four to seven MLN per line were randomly selected, with the maximum circumference of each measured using a flexible measuring tape by the primary author or a technician, and entire length incised. Two 15 mm cross-sections were taken from each LN and stored in two individual sterile containers, one with 10% buffered formalin. Samples were collected and processed for culture and histopathology as above.

5.3.2.2 Data collection and management

Carcass weight, age, sex and herd location were recorded at the time of sampling. Age was categorised as young or adult as above. The histopathological features of follicular hyperplasia, capsular infiltration by eosinophils, foci of macrophages containing lipopigments, parasitic granulomas, calcified foci, focal granulomas and/or reaction to a ZN stain were categorised, graded and coded for later analysis as outlined in Chapter 7. Mesenteric LN circumference, all histopathological features, age, sex, carcass weight and herd were recorded at the carcass level.

5.3.2.3 Statistical analysis

Mesenteric LN circumference was categorised into 5 mm increments, starting with those 35 mm and above, and used for determination of a cut point. The specificity (Sp) and sensitivity (Se) of MAP diagnosis were plotted as cumulative percentages at each MLN circumference increment. The Se was defined as the correct diagnosis of MAP infection within a MLN truly culture- positive for the bacteria, whereas the Sp was defined as the correct diagnosis of no MAP infection within a MLN truly culture-negative for the bactiera. The circumference cut point of

108 an ‘abnormal’ MLN was defined as the lowest MLN circumference with a Sp of MAP diagnosis >95%.

Potential factors associated with a MAP culture-positive MLN, such as carcass weight, age, sex and histopathological features, were screened for association with MLN circumference. Factors associated with MLN circumference with a p-value of <0.2 were included in a multivariable, linear mixed-effects model. Three MLN were removed from the analysis as animal age was unavailable. In order to correct for a skewed distribution of residuals, the outcome variable was log transformed. Herd of origin was included in the model as a random effect to adjust variance estimates for correlation at this level of aggregation. Age was forced into the final model to control for confounding. Biologically plausible interactions between fixed effects were tested for significance and model fit and residuals were evaluated for distributional assumptions and influential observations. A boxplot was used to show the mean and scatter of circumference in MLN positive and negative for MAP, stratified by the presence of focal granulomas. Four outlying MLN circumference measurements (culture negative MLN >90 mm circumference containing focal granulomas) were removed and the model re-run to assess their impact on the complete model. All analyses were performed in R (version 2.4.1; The R Foundation for Statistical Computing) and significance was declared at p < 0.1.