Abstract
In this study, ninety one blood samples were collected from unvaccinated cattle with a history of Brucella melitensis biovar 3 infection and belonged to the Nile Delta region during slaughter in abattoirs. Serum samples were separated and sample size were estimated according to OIE requirements (2% errors and 95% confidence interval) for the validation of two developed iELISA techniques using different antigens namely lipopolysaccharide antigen (LPS) and soluble periplasmic protein antigen (PPP) against an iELISA (LPS) commercial kit. Diagnostic performance characteristics were estimated considering CFT as the gold standard. The estimated relative sensitivities (Se) and specificities (Sp) of the three iELISA versions were as follows: iELISA (PPP) (88% and 94%), iELISA (LPS) (92% and 87%) commercial kit and home-made iELISA (LPS) (90% and 84%). The estimated κ agreement values with the CFT indicated substantial agreement in case of iELISA versions used both LPS and PPP antigens. The performance of iELISA based on both ROCs and AUCs was very good being equal to or closer to 0.9. Performance indices (PI) values were charted in ascending order. The accuracy % was calculated to be 90% for both iELISA (PPP) and iELISA (LPS) kit and 88% for home-made iELISA (LPS). Highest positive predictive values were achieved by iELISA (PPP), while the highest negative predictive values were achieved by iELISA (LPS) standard kit. The highest positive likelihood ratio (LR+) was attained by iELISA (PPP), while the negative likelihood ratio (LR-) was almost the same. Authors suggest that the home-made iELISA (LPS) is fit for its purpose as a rapid screening test. The high performance characteristics of iELISA (PPP), especially the balance of both Se and Sp, necessitates a reclassification of this version as confirmatory rather than screening test with further studies.

Key words: brucellosis,ELISA,diagnosis,validation,bovine.