Insights from Alberta youngstock testing strategies and estimation of test accuracy using Bayesian latent class analysis to evaluate Mycobacterium avium subspecies paratuberculosis (MAP) infection in Canadian dairy cattle

Johne’s disease (JD) is the clinical manifestation of the infection of Mycobacterium avium subspecies paratuberculosis (MAP). JD remains difficult to control due to prolonged subclinical infection, intermittent fecal shedding, and the absence of a diagnostic gold standard, due to the limitations of currently available diagnostic tests. Most JD control programs focus exclusively on adult cattle, thereby overlooking the role of youngstock in MAP transmission, despite evidence that they may contribute to within-herd transmission long before they are actively included in JD testing strategies. The objectives of this thesis were to (i) determine the age at which youngstock ≤12 months of age in MAP-infected dairy herds begin shedding MAP in feces and developing detectable antibody responses under field conditions, and (ii) estimate the sensitivity and specificity of serum ELISA, individual fecal qPCR, and individual fecal culture for MAP detection in adult dairy cows using a Bayesian latent class model accounting for conditional dependence between tests. In Chapter 2, youngstock (animals ≤12 months of age) were tested from eight MAP-infected dairy herds in Alberta, Canada, demonstrating that MAP shedding occurred in 13% of animals ≤12 months of age, with shedding detected as early as four months of age. Serological responses were detected in 4% of youngstock, with ELISA positivity observed from three months of age. These results provide field-based evidence that youngstock can shed MAP and develop immune responses earlier than previously assumed, potentially contributing to within-herd MAP transmission. In Chapter 3, the performance of three commonly used diagnostic tests for MAP detection: serum ELISA, fecal qPCR, and individual fecal culture was evaluated. Blood and fecal samples were collected from adult dairy cows in MAP-infected herds in Quebec, Canada. The Bayesian latent class model framework was used to estimate the sensitivity (Se) and specificity (Sp) of each diagnostic test in the absence of a gold standard. Results yielded a Se of 52%, 44% 92% respectively for culture, ELISA and qPCR, and specificities of 99%, 96%, 88%, respectively. Fecal culture and serum ELISA exhibited moderate Se but high Sp, whereas fecal qPCR demonstrated substantially higher Se with slightly lower Sp, demonstrating its ability to detect shedding animals, suggesting that qPCR can enhance early identification of infected cattle.