Abstract
Salmonid alphavirus (SAV) is the OIE-listed, viral cause of pancreas disease (PD) in farmed Atlantic salmon. SAV is routinely detected by PCR-methods while typical histopathological lesions are additionally used to confirm the diagnosis. Field evaluation of diagnostic test performance is essential to ensure confidence in a test's ability to predict the infection or disease status of a target animal. For most tests used in aquaculture, characteristics like sensitivity (Se) and specificity (Sp) at the analytical level may be known. Few tests are, however, evaluated at the diagnostic level according to the OIE standard. In the present work, we estimated diagnostic test sensitivity (DSe) and diagnostic test specificity (DSp) for five laboratory tests used for SAV detection. As there is no gold standard, the study was designed using Bayesian latent class analysis. Real-time RT-PCR, cell culture, histopathology, virus neutralization test, and immunohistochemistry were compared using samples taken from three different farmed Atlantic salmon populations with different infection status; one population regarded negative, one in an early stage of infection, and one in a later stage of infection. The average fish weight in the three populations was 2.0, 1.6, and 1.5 kg, respectively. The DSe and DSp of real-time RT-PCR is of particular interest due to its common use as a screening tool. The method showed high DSe (≥0.977) and moderate DSp (0.831) in all 3-populations models. The results further suggest that a follow-up test of serum samples in real-time RT-PCR negative populations may be prudent in cases where epidemiological information suggest a high risk of infection and where a false negative result is of high consequence. This study underlines the need to choose a test appropriate for the purpose of the testing. In the case of a weak positive PCR-result, a follow-up test should be conducted to verify the presence of SAV. Cell culture showed high DSe and DSp and may be used to verify viral presence.