Validation of a Neospora caninum iscom ELISA without a gold standard

Neospora caninum is an intracellular parasite which causes abortion in cattle worldwide. One problem in the validation of the different methods for demonstration of this parasite is the lack of an appropriate gold standard. To validate an immunostimulating complex (iscom) enzyme-linked immunoassay (ELISA) used to detect antibodies to N. caninum, sera from 244 cattle in five Swedish dairy herds infected with N. caninum were analysed. The sera also were analysed by a standard indirect-fluorescent antibody test (IFAT). The results obtained by the two tests were compared using the Gibbs sampler. Gibbs sampling is a latent-class approach based on Bayesian statistics; neither test is assumed to be more correct in stating the true status of infection. The Gibbs sampler was run using both informative and non-informative prior probabilities. We also simulated different cut-offs in the iscom ELISA (providing data to inform selection of optimal cut-off values for different applications). The ELISA produced fewest incorrect test results over all at a cut-off value of 0.200. The sensitivity and specificity at this cut-off were 99 and 96%, respectively. The IFAT had a high specificity (99%) but a lower sensitivity (78%) than expected-confirming that the IFAT cannot be treated as a true gold standard. Sensitivity and specificity of the ELISA were presented in a two-graph receiver operating characteristic (TG-ROC) plot. Any cut-off between 0.150 and 0.300 will have both sensitivity and specificity > or =95%. Optical densities of < or =0.150 and > or =0.550 (or > or =0.350) were suggested as limits to rule out and rule in infection, respectively.