Diagnostic accuracy of three commercial immunoenzymatic assays for small ruminant lentivirus infection in goats performed on individual milk samples.

ABSTRACT

Caprine arthritis-encephalitis (CAE) caused by small ruminant lentivirus (SRLV) infection is one of the most widespread and devastating diseases of goats. Serological methods, mainly immunoenzymatic assays (ELISA), are the mainstay of CAE diagnostics. Even though blood is still the most commonly tested material, animal welfare issues and increasing costs of veterinary service prompt the development of serological methods based on milk testing. Several different types of ELISAs for CAE are available on the market. All of them perform well on serum, however their diagnostic accuracy for testing milk has not been so far compared. Therefore, we carried out the study in 5 dairy goat herds in Poland whose previous epidemiological situation regarding CAE was known. Paired serum and milk samples were collected from all adult females (n = 420) and tested with 3 commercial ELISAs - indirect ELISA based on the whole-virus antigen (wELISA), indirect ELISA based on the recombined transmembrane and capsid protein (TM/CA-ELISA), and competitive ELISA based on the surface glycoprotein (SU-ELISA). Milk was tested as lactoserum at dilution of 1/2 in wELISA and TM/CA-ELISA, and undiluted in SU-ELISA. The true status of goats was based on the composite reference standard comprising the results of all three ELISAs done on serum and the true prevalence of SRLV infection in the herd of origin. 243 (57.9 %) goats were classified as truly positive and 177 (42.1 %) goats as truly negative. Diagnostic accuracy was evaluated using the area under the ROC curve (AUROC) as well as sensitivity (Se) and specificity (Sp) for a range of cut-off values. AUROC was 98.8 % (CI 95 % 97.5 %, 100 %) for wELISA, 97.9 % (CI 95 % 96.5 %, 99.2 %) for TM/CA-ELISA, and 91.7 % (CI 95 % 88.9 %, 94.5 %) for SU-ELISA. At the cut-off values recommended by the manufacturers both indirect ELISAs were highly sensitive (89.3 % and 91.4 %, respectively) and highly specific (98.3 % and 95.5 %, respectively), whereas SU-ELISA had only moderate Se (71.2 %) at comparably high Sp (96.6 %). Nevertheless, the optimal cut-off values were lower than those recommended by manufacturers for serum - sample-to-positive control serum ratio (S/P%) of 10 % for wELISA, S/P% of 80 % for TM/CA-ELISA, and percentage inhibition of 23 % for SU-ELISA. Concluding, the study shows that wELISA and TM/CA-ELISA may be interchangeably used for testing individual goat milk samples for SRLV infection. Diagnostic sensitivity and specificity of these ELISAs appear not to be lower on milk than on serum. SU-ELISA is considerably less sensitive on milk samples than indirect ELISAs.