M108

The objective of this study was to validate a new blocking enzyme-linked immunosorbent assay (ELISA) (designated M108 for milk and S108 for serum samples) for detecting bovine leukemia virus (BLV) infection in dairy cattle. Milk, serum, and ethylenediaminetetraacetic acid–blood samples were collected from 524 adult Holstein cows originating from 6 dairy herds in Central Argentina. The M108 and S108 were compared with agar gel immunodiffusion (AGID), polymerase chain reaction and a commercial ELISA. Because there is currently no reference test capable of serving as a gold standard, the test sensitivity (SE) and specificity (SP) were evaluated by the use of a latent class model. Statistical inference was performed by classical maximum likelihood and by Bayesian techniques. The maximum-likelihood analysis was performed assuming conditional independence of tests, whereas the Bayesian approach allowed for conditional dependence. No clear conclusion could be drawn about conditional dependence of tests. Results with maximum likelihood (under conditional independence) and posterior Bayes (under conditional dependence) were practically the same. Conservative estimates of SE and SP (with 95% confidence intervals) for M108 were 98.6 (96.7; 99.6) and 96.7 (92.9; 98.8) and for S108 99.5 (98.2; 99.9) and 95.4 (90.9; 98.1), respectively. The ELISA 108 using either milk or serum to detect BLV-infected animals had comparable SE and SP with the official AGID and a commercial ELISA test, which are currently the most widely accepted tests for the serological diagnosis of BLV infection. Therefore, ELISA 108 can be used as an alternative test in monitoring and control programs.