[The suitability of ELISA for the detection of antibodies against Mycobacterium avium ssp. paratuberculosis in bulk milk samples from Rhineland-Palatinate].

Paratuberculosis or Johne's Disease, caused by Mycobacterium avium subspecies paratuberculosis (MAP), is a notifiable disease in Germany which produces enormous economical losses in dairy farms. At present,there is no confirmed data about the actual number of infected livestock herds in Germany. A countrywide monitoring program to evaluate the prevalence in dairy herds would only be economically feasible on the basis of bulk milk testing. In this study, we evaluated two ELISA test kits (SVANOVIR Ptb-ELISA, IDEXX-M.pt. Milk test kit) for the detection of antibodies against MAP in bulk milk. First, the Paratuberculosis-status of the herd derived from the history of the farm was used as a gold standard. Paratuberculosis-negative farms were tested negative with each test, but paratuberculosis-positive or Paratuberculosis-serologically-positive farms were detected only in one case (Svanovir) or three cases (IDEXX), respectively. Even if inconclusive results are counted as positive, 82.9 % (Svanovir) or 80 % (IDEXX) of the paratuberculosis-positive or serologically paratuberculosis positive farms were not detected. Nevertheless, a re-validation of both ELISAs by means of ROC and TG-ROC analyses was attempted by searching for ideal cut-offs, optimised for bulk milk. If a high specificity was selected, no acceptable sensitivity could be reached. The best results were obtained using a sensitivity of 32.3 % at a specificity of 100 % (Svanovir). With a small change of the cut-off value, the sensitivity increased to still 57 %, but this reduced the specificity to 67 %. Similar results were obtained with the IDEXX-ELISA. We then evaluated the Svanovir-ELISA for the detection of bulk milk samples on the basis of the current paratuberculosis prevalence within 69 dairy herds from Rhineland-Palatinate using individual milk samples.When the bulk milk samples were tested in two different laboratories using the same ELISA, considerable differences in the results became evident. Nearly all samples were tested with a higher relative test result in one laboratory, which often led to differences in the classification of the prevalence levels. The estimated within-herd seroprevalences ranged between 0 % and 37 %. There was little agreement between the historical paratuberculosis herd status and the within-herd prevalence in milk serum, as reflected in a kappa-index of 0.146. To determine the sensitivity and specificity of the bulk milk ELISA by ROC and TG-ROC analysis, 116 bulk milk samples were used that had been obtained from the 69 dairy herds participating in the study. The optimal ratio of sensitivity (81 %) and specificity (77 %) relative to a "gold standard" was obtained when the cut-off was set at the 10 % level. These values for sensitivity and specificity were better than those obtained in an evaluation of the same ELISA in which the historical Paratuberculosis herd-status was used as a "gold standard." The results of this study question the suitability of the available ELISAs for bulk milk testing. Taking into account that the Svanovir-ELISA for individual milk samples has a sensitivity of 60 96% relative to the blood serum variant of the test, and that the latter has also a limited sensitivity due to the pathogenesis of paratuberculosis, the available test systems examined in this Study do not seem to be suitable for herd diagnosis by using bulk milk samples.