Validation of an indirect enzyme-linked immunosorbent assay for the detection of antibody against Brucella abortus in cattle sera using an automated ELISA workstation.

An automated indirect enzyme-linked immunosorbent assay (I-ELISA) for the serological diagnosis of bovine brucellosis was developed and validated in-house. A total of 4,803 cattle sera from South Africa (n = 3,643), Canada (n = 652), Germany (n = 240), France (n = 73) and the USA (n = 195) was used. The South African panel of sera represented 834 sera known to be positive by the Rose Bengal test (RBT), serum agglutination test (SAT) and complement fixation test (CFT), 2709 sera that were negative by CFT, and 100 sera from animals vaccinated with a standard dose of Brucella abortus strain 19. Overseas sera were obtained from reference non-vaccinated brucella-free cattle (n = 834), naturally infected (n = 72), experimentally infected (n = 71), and vaccinated animals (n = 83). Also 100 sera collected from cattle in Canada and known to be positive by competitive ELISA (C-ELISA) were used. The intermediate ranges ("borderline" range for the interpretation of test results) were derived from two-graph receiver operating characteristics analysis. The lowest values of the misclassification cost-term analysis obtained from testing overseas panels, covered lower I-ELISA cut-off PP values (0.02-3.0) than those from local panels (1.5-5.0). The relatively low cut-off PP values selected for I-ELISA were due to the fact that the positive control used represents a very strong standard compared to other reference positive sera. The greater overlap found between negative and positive cattle sera from South Africa than that between reference overseas panels was probably due to the different criteria used in classifying these panels as negative (sera from true non-diseased/non-infected animals) or positive (sera from true diseased/infected animals). The diagnostic sensitivity of the I-ELISA (at the optimum cut-off value) was 100% and of the CFT 83.3%. The diagnostic specificity of I-ELISA was 99.8% and of the CFT 100%. Estimate of Youden's index was higher for the I-ELISA (0.998) than that for the CFT (0.833). Analysis of distribution of PP values in sera from vaccinated and naturally infected cattle shows that in vaccinated animals all readings were below 31 PP where in infected ones these values represented 43%. Therefore, it appears that I-ELISA could be of use in identifying some naturally infected animals (with values > 31 PP), but more sera from reference vaccinated and infected animals need to be tested to further substantiate this statistically. Of 834 sera positive by RBT, SAT and CFT, 825 (98.9%) were positive in the I-ELISA. Compared to C-ELISA the relative diagnostic sensitivity of the I-ELISA was 94% and of the CFT 88% when testing 100 Canadian cattle sera. Of 258 South African cattle sera, of which 183 (70.9 %) were positive by the I-ELISA and 148 (57.4 %) by the CFT, 197 (76.4%) were positive by C-ELISA when re-tested in Canada. One has to stress, however, that Canadian C-ELISA has not been optimised locally. Thus, the C-ELISA was probably not used at the best diagnostic threshold for testing South African cattle sera. This study shows that the I-ELISA performed on an automated ELISA workstation provides a rapid, simple, highly sensitive and specific diagnostic system for large-scale detection of antibodies against B. abortus. Based on the diagnostic accuracy of this assay reported here, the authors suggest that it could replace not only the currently used confirmatory CFT test, but also the two in-use screening tests, namely the RBT and SAT.