Assessment of two enzyme-linked immunosorbent assay tests marketed for detection of ruminant proteins in finished feed.

The performance characteristics of two enzyme-linked immunosorbent assay (ELISA) test kits, ELISA Technologies' MELISA-Tek test and Tepnel BioSystems' BioKit for (Cooked) Species Identification test, designed to detect ruminant proteins in animal feed, were evaluated. The test kits were evaluated by using acceptance criteria developed by the U.S. Food and Drug Administration's Center for Veterinary Medicine Office of Research for evaluating selectivity, sensitivity, ruggedness, and specificity. The acceptance criteria for determining success used a statistical approach requiring a 90% probability of achieving the correct response within a 95% confidence interval. In practice, this measure requires the test to achieve the correct response 58 times for every 60 samples evaluated, or a 96.7% accuracy rate. A minimum detection level of 0.1% bovine meat and bone meal (BMBM) was required, consistent with the sensitivity of the analytical methods presently used by the U.S. Food and Drug Administration. Selectivity was assessed by testing 60 dairy feed samples that contained no added animal proteins; sensitivity was determined by evaluating 60 samples (per level of fortification) of this same feed that contained 0.025, 0.05, 0.1, 0.25, 0.5, 1, or 2% BMBM. The MELISA-Tek test passed the acceptance set-point criteria for selectivity assessment but failed the sensitivity assessment at all levels except at the 2% level. The MELISA-Tek test came close to passing at the 1% level, detecting true-positive findings at a rate of 93%, but failed at lower levels, in spite of the label claim of 0.5% sensitivity. The BioKit for (Cooked) Species Identification test detected only 2 of 17 samples fortified at the 2% BMBM level and failed to detect any other BMBM-fortified samples. The results of this evaluation indicate that neither test is adequate for regulatory use.

Evaluation of two commercial lateral-flow test kits for detection of animal proteins in animal feed.

Performance characteristics were evaluated for two lateral-flow test kits, Reveal for Ruminant in Feed (Neogen Corporation) and FeedChek (Strategic Diagnostics Inc.), designed to detect ruminant or terrestrial animal proteins in feeds. The stringent acceptance criteria used were developed by the Center for Veterinary Medicine Office of Research to identify test kits with comparable selectivity and sensitivity to microscopy and PCR assay, the analytical methods used by the U.S. Food and Drug Administration (FDA). Guidelines were developed for evaluating the selectivity, sensitivity, ruggedness, and specificity of these kits. These guidelines further stated that ruggedness and specificity testing would be performed only after a test passed both the selectivity and sensitivity assessments. Acceptance criteria for determining success were developed using a statistical approach requiring 90% probability of achieving the correct response, within a 95% confidence interval. A minimum detection level of 0.1% bovine meat and bone meal, consistent with the sensitivity of the methods used by the FDA, was required. Selectivity was assessed by testing 60 dairy feed samples that contained no added animal proteins; sensitivity was determined by evaluating 60 samples (per level of fortification) of the same feed that contained 0.025, 0.05, 0.1, 0.25, 0.5, 1, or 2% bovine meat and bone meal. The Reveal test passed the selectivity assessment but failed the sensitivity assessment, detecting only samples fortified at the 2% level and then only 17 to 33% of those samples, when read according to the label directions. The FeedChek test passed the sensitivity assessment but failed the selectivity assessment, with rates for false-positive results ranging from 34 to 38%, depending on the user. The sensitivity of the Reveal test was affected by the concentration of trace minerals present in the feed; concentrations toward the high end of the normal range prevented the detection of true positive feed samples containing bovine meat and bone meal. Better sensitivity assessments were obtained when lamb meal was used either alone or in combination with bovine meat and bone meal. The FeedChek test was not affected by the concentration of trace minerals or by the type of animal meal used. These results indicate that neither of the two tests is adequate for routine regulatory use.