Background.

Several serological tests (ELISA, CFT, AGIDT) are available for the detection of antibodies against M. paratuberculosis. For the detection of antibodies in cattle ELISAs are most widely used. The reproducibility and sensitivity/specificity of ELISAs are considered better than those of alternative tests. Additionally, ELISAs are less laborious and therefore allow high throughput testing. Herd sensitivity and herd specificity are important criteria for certification programs. To calculate these parameters, it is of utmost importance to determine test characteristics like sensitivity, specificity and reproducibility accurately. Therefore, we performed an extensive comparative validation study of five commercially available ELISAs.

Methods.

Sensitivity, specificity, diagnostic performance, and reproducibility of commercially available ELISAs for detection of antibodies against M. paratuberculosis were evaluated using different serum panels from Dutch, American and French cattle populations. Relative sensitivity was evaluated taking faecal shedding as a gold standard (288 sera from faecal shedders from different origin in total). Specificity was evaluated using in total 2135 sera originating from repetitively faecal culture-negative herds from different origin. Diagnostic performance of tests was evaluated by ROC analysis, and effects on sensitivity and specificity of parallel testing schemes were evaluated. Additionally, detection limit, reproducibility (within and between plate variance) and other technical criteria were evaluated.

Results.

Results are too extensive to present in this abstract. However, clear differences in overall sensitivity, specificity and diagnostic performance were found between ELISAs. Also, for subpanels of different origin differences in specificity were found, possibly reflecting regionally relevant cross-reactions. Results of four out of five ELISAs were strongly correlated, whereas results of one ELISA showed low correlation coefficients with all other tests. Parallel testing using the latter ELISA with one of the other ELISAs would theoretically yield a higher combined sensitivity but a lower combined specificity. The feasibility of such a testing scheme, based on the underlying data, will be discussed besides the other parameters investigated.

Conclusions.

Most of the commercial ELISAs evaluated had similar overall accuracy but important differences were observed that must be considered when selecting the best ELISA kit for a laboratory.