The assessment of field trials in GMO research around the world and their possible integration in field trials for variety registration

Most regulations worldwide stipulate that a new genetically modified (GM) crop event has to be compared to its closest non-GM counterpart as a corner stone of the pre-market risk assessment. To this end the GM crop and its comparator should be grown in field trials for a phenotypic comparison as well as for subsequent detailed analysis of the composition of the two crop varieties. A more in-depth globally harmonised approach for the conduct of these field trials is lacking. Only a few countries have formulated detailed protocols for the set-up of GM field trials. In some countries, commercial non-GM reference varieties need to be included in a field study to compile reliable data that indicate the range of natural variation for the compounds tested at the specific location. Detailed analysis of pre-market assessment reports have so far not shown the added value of including these reference varieties in the field trials. In all cases where specific values were found to be outside of the range of the reference varieties, it proved possible to draw conclusions on the part of the pre-market risk assessment that relates to the compositional analysis, on the basis of already available compositional data. With the increasing quality of several databases on compositional data of a growing number of crop species, it seems unlikely that reference varieties will become more important on future occasions. It was furthermore investigated whether this part of the risk assessment can be related to field trial requirements for variety registration with the explicit intention of reducing the data burden on producers of new GM plant varieties. Field trials for variety registration so far include an assessment of phenotypic characteristics that do not cover safety aspects, with the exception of establishment of the glycoalkaloid content in potatoes in the Netherlands and Sweden. It may, however, under certain conditions be relatively easy to exchange data from compositional measuremen