EU Science Hub

EURL GMFF Proficiency Test Report CT 02/17: Determination of GM Soybean 40-3-2 in Chicken Feed and Soybean Flour

The European Union Reference Laboratory for Genetically Modified Food and Feed (EURL GMFF) organised a proficiency test (PT) for National Reference Laboratories (NRLs) to support the official controls on food and feed in line with Regulation (EC) No 882/2004. Other official control laboratories were allowed to participate on a voluntary basis. Two test items were distributed: test item 1 (T1) was composed of ground chicken feed spiked with a mixture of GM soybean event MON-Ø4Ø32-6 (40-3-2) and non-GM soybean, and test item 2 (T2) was a soybean flour containing the same GM soybean event 40-3-2. Participants were required to screen T1 and T2 for the presence of three GM soybean events, and to quantify the event that was present with the highest GM mass fraction. The results had to be reported in GM mass fraction (mass/mass %). Eighty-six participants from 39 countries participated to this PT, including 54 NRLs, of which 33 are designated in line with Regulation (EC) No 882/2004 (NRL/882) and 21 are nominated in Regulation (EU) No 120/2014 to support the EURL GMFF on method validation (NRL/120), as well as 10 other EU official control laboratories. The qualitative results, i.e. the correct identification of the GM event, were evaluated and scored as correct or incorrect. The assigned value for the 40-3-2 soybean mass fraction in both test materials was derived as the robust mean of the data provided by the NRLs. Laboratory performance was primarily evaluated by calculating z scores. The results reported indicate that all participants identified the correct GM event in both test items. All, but one NRL obtained a satisfactory performance (z) score for the reported 40-3-2 soybean mass fraction in both test items and the performance of one other NRL was unacceptable for T1 because they reported the 40-3-2 mass fraction as below their LOQ. Six and two NRLs obtained a questionable z score for T1 and T2, respectively. Considering the results provided by the other participants (non-NRL), three and two non-NRLs obtained an unsatisfactory z score for T1 and T2, respectively. Another three laboratories obtained a questionable z score for T1 and one laboratory for T2. The laboratories' ability to provide results close to the assigned value within their claimed measurement uncertainty was additionally evaluated by ζ scores. Twenty four and 16 laboratories had an unsatisfactory ζ scores for T1 and T2, respectively. Unsatisfactory ζ scores were mainly the consequence of an underestimated or not reported measurement uncertainty. Guidance is provided for correctly estimating the measurement uncertainty of analytical results. A root-cause analysis will be requested from NRLs with an unsatisfactory z score result in this PT and will be followed-up.