Comparison of enzyme immunoassays for the diagnosis of bovine brucellosis

Abstract

The indirect enzyme immunoassay for measurement of bovine antibody to Brucella abortus was tested on 15 716 Canadian sera to assess the specificity. These sera were also tested by the buffered plate antigen test. Two enzyme-linked immunosorbent assay (ELISA) formats were used for assessment of data: the targeting procedure using a positive control serum allowed to develop to an optical density of 1.0 and the use of a positive control serum to determine relative positivity at a set time. Two different cut-off values were also assessed for each assay. A total of 763 sera gave reactions above established cut-off values in the ELISA while 216 were positive in the buffered plate antigen test (BPAT). A modification of the indirect ELISA employed divalent cation chelating agents (EDTA/EGTA) incorporated into the serum incubation stage to eliminate some non-specific reactions. This method was applied only to the 763 indirect ELISA reactor sera and it eliminated all but 93 or 37, depending on the cut-off selected, of the reactions. Sensitivity was assessed by testing 424 sera from B. abortus culture positive cattle. The indirect ELISA classified all 424 sera as positive by either method of data handling and with or without addition of EDTA/EGTA for a specificity estimate of 100%. In the BPAT, 412 sera gave a positive agglutination reaction. Ten percent of the 15 716 sera were randomly selected and tested by two different competitive ELISAs and by the complement fixation test. One competitive ELISA used B. abortus O-polysaccha-ride as the antigen and an enzyme conjugated monoclonal antibody to the O-polysaccharide for competition and detection. Of the sera tested, 34 gave false positive reactions. On a retest, the false positive reactions were reduced to two. The second competitive ELISA used lipopolysaccharide as the antigen, a different monoclonal antibody but also specific for the O-polysaccharide for competition and commercially available goat anti-mouse IgG enzyme conjugate for detection. In the initial assessment, this test gave rise to five false positive reactions. This number was reduced to two when retesting the sera. The complement fixation test used was a micro format test and three sera gave false positive reactions. A total of 654 sera from animals from which B. abortus was isolated was tested by the first competitive ELISA. Of these, nine sera were negative on the initial test. This number was reduced to three on repeat testing. All 636 sera tested by the second competitive ELISA were positive. Fifteen of 636 sera gave sufficient prozoning in the complement fixation test to be considered diagnostically negative and 59 sera were anticomplementary. Sensitivity and specificity are summarized in the table. Cut-off values for each assay were initially established by visual observation of frequency distributions of positive and negative serum samples. These cut-off values were confirmed by receiver operating characteristics (ROC) analysis. In addition, an index of performance ( accuracy ) was established for each assay in order to allow direct comparison. Accuracy estimates were based on the sensitivity, specificity and disease prevalence for the data.