Adaptation of a surface antigen-based ELISA for the detection of antibodies against Neospora caninum in bovine milk

The aim of the present study was to determine whether an ELISA based on a 38-kDa surface antigen (NcSRS2) of Neospora caninum tachyzoites could be used to examine bovine milk for antibodies against N. caninum. A total of 797 undiluted milk samples from N. caninum-infected herds were examined in the p38-ELISA. Milk results of individual animals were compared with those obtained by the same ELISA for the corresponding serum samples. The linear correlation between milk and serum antibody results of individual animals was characterized by an adjusted R-2 of 0.644. The examination of the two-graph receiver-operating characteristics revealed an optimal cut-off of 0.150 to obtain similar results in the examination of milk and serum. With this cut-off, the test had a specificity and a sensitivity relative to the serum results of 93%. Using this cut-off, excellent agreement was observed between milk and serum results (Kappa 0.85; 95% CI, 0.781-0.920). A cut-off of 0.450 resulted in a specificity of 99% relative to the serum results. At this cut-off, the sensitivity of the test was 80% relative to the serum-ELISA and agreement was slightly lower (Kappa = 0.82; 95% CI, 0.749-0.887). An optimized linear regression model suggests that, in addition to the result in the p38-serum-ELISA, the variables abortion risk (ABRISK) (abortion between 100 and 269 days pregnant) and the age of the animal (AGE) affected the result of the p38-milk-ELISA. The optimized linear regression model had an adjusted R-2 of 0.8501