Improvement in the diagnosis of Brucella abortus infections in naturally infected water buffaloes (Bubalus bubalis) using an ELISA with a Protein-G-based indicator system

Brucellosis caused by Brucella abortus in domestic water buffaloes (Bubalus bubalis) raised under the traditional system of husbandry in northern India was diagnosed using an enzyme-linked immunosorbent assays (ELISA) with a Protein-G-based indicator system (Protein-G ELISA). A total of 1,551 animals that are positive (N = 61), negative (N = 243), and suspected (N = 1,247) for brucellosis were examined. Rose bengal test (RBT) was used to predict the disease, and accordingly, animals were dichotomized in positive and negative population for receiver operating characteristic (ROC) curve analysis to determine the sensitivity, the specificity, and the performance index of Protein-G ELISA. Taking all animals (N=1551) into account, the ROC curve analysis revealed cut off value of 29.6% positivity (%P) with 98.40% and 94.94%, sensitivity and specificity, respectively. The results were compared with ELISA in which anti-bovine conjugate was used. The cut off in ELISA was 37.9%P and sensitivity and specificity were 96.26% and 97.07%, respectively. The performance indexes of both the assays were almost equal and were 193.34 for Protein-G ELISA and 193.33 for ELISA. The cut off values of both the tests changed, if only known positive (N = 61) and known negative (N=243) animals were used for ROC curve analysis, and accordingly, changes in sensitivity and specificity were observed with significant decrease of performance indexes of both the tests. The high optical density (P<0.0001) background signal with negative serum control and high %P (P<0.0001) in sera from negative population were noticed in ELISA in comparison to Protein-G ELISA.