Performance assessment under field conditions of a rapid immunological test for transgenic soybeans

Current market conditions and food regulations make it necessary for international and domestic participants in the agrifood industry to structure supply chains that control the content of genetically modified (GM) material in their products. Tests to detect and/or quantify GM components represent an important tool in maintaining such supply systems. This study assesses the field performance of kits that employ lateral flow immunotechnology to detect soybeans GM to be resistant to the herbicide glyphosate. Operators at 23 grain-handling facilities were paid to conduct analyses on a series of blinded samples containing defined proportions of conventional and transgenic soybeans. The observed rate of false positives was 6.7% in an experiment in which the highest level of GM material was 1% and 22.3% in a second experiment in which the highest level of GM material was 10%. This difference may be attributed to increased risk of cross-contamination with the higher level of transgenic material used in the second study. Samples containing 0.01% GM material were reported as genetically modified 6.70% of the time, while samples containing 0.1, 0.5 and 1% GM material were classified as genetically modified 29.5, 67.7 and 68.2% of the time, respectively. Thus, the frequencies of false negatives were 93.3, 70.5, 33.3 and 31.8% for samples containing 0.01, 0.1, 0.5 and 1.0% GM material. Samples containing 10% GM material were correctly reported as genetically modified in all cases. These results lead to the conclusion that the kit under study is useful in screening for lots of soybeans that contain high levels of GM material, but that, as a field tool, it is not effective in monitoring for GM material at the level of 1.0% or lower. Statistical and immunochemical analyses were carried out in order to assess the relative contributions of various factors to the error observed in these studies. These analyses indicated that limitations in operator performance, not defects in test kit materials, were the primary contributors, while sample size may play a secondary role. As both operator performance and sample size are independent of the specific characteristics of the test kit used in this study, it appears justifiable to generalize conclusions obtained here to other similar test systems.