GMO traceability: “Simplified and harmonised sampling procedures are essential”
The main reason why sampling is so critical is that GMOs are very often unevenly distributed in lots and in products. Therefore, it is difficult to produce a sample that is representative for the entire lot. A representative sample is the starting point for any kind of testing. If sampling is done improperly, the results of even the most carefully conducted GMO test can be misleading. This would undermine the integrity of GMO labelling, which as you know is required in the EU. Improper sampling can thus mislead both food producers and consumers about the GM-content of products. Improper sampling could even have impacts on public health. If GMO tests fail to detect an untested or unauthorised GM crop, there could be potential risks to human and animal health.
The probability of getting a false positive due to improper sampling is extremely low. If a result is positive, there is a much higher probability that the majority of the lot is GM. Here’s an analogy: imagine you have a large box with many small black balls and only a few white ones. If you take just a handful of balls, you are more likely to have only black balls in your hand than you are to have at least one white one. With the actual content of the box unknown, getting a white ball means there are two possibilities. You were either very lucky to get the rare white ball, or the box has a lot of white balls. Therefore, if you do get a positive, it is likely that there is widespread GM content. The outcome of poor sampling would most often be a false negative.
Actually, at least in Europe, there are very detailed guidelines, like EU recommendation 707/2004, that provide very reliable and detailed guidelines for sampling in bulk. In fact, if stakeholders simply ignore costs and follow the recommendations as stated, reliability should be sufficiently high. The problem with the current guidelines is that they are very detailed and involve a large number of incremental samples. This is ultimately very costly, and in some cases such extensive sampling is not even necessary. For instance, at some points along the food processing chain it could be easier and just as reliable to perform more simplified sampling procedures. Guidelines for these different possibilities have not yet been specified, because the existing guidelines were designed for the testing of grains.
Powdered material and pastes are easier to analyze than grains. The more finely ground the material, the more it behaves like a liquid. For GMO testing, this means that fewer incremental samples are needed. Of course, this must be studied on a case-by-case basis depending on the processing chain. It is not very easy to provide generalised recommendations when there are so many different types of products and materials.
Another possibility is sampling at retail points. Laws in Europe require labelling for products containing GMOs, and the emphasis, from the point of view of the consumer, is on the final product. It is the duty of control authorities to sample products as they are sold. This is to ensure that labelling is in compliance with regulations. Although the existing recommendations on sampling do take into account sampling for packaged products, there are no clear, well-defined guidelines. My group is conducting studies on this complex topic in the framework of Co-Extra, but much still remains to be accomplished. We hope that these studies will yield pragmatic guidelines that provide assistance on a case-by-case basis.
There are software tools that can aid in developing sampling plans for a various situations. Many software tools exist, but they are not integrated into a platform that is easy for stakeholders to access and use. Currently, sampling software does not exist as a ready-to-use instrument for stakeholders. Other groups in Co-Extra are carrying out studies to develop a more user-friendly, integrated platform such as the DSS (Decision Support System).
There is in fact an automatic sampling device that can greatly improve the feasibility of sampling according to the current recommendation. At the moment, many operators that carry out sampling do not have this instrument at their disposal. It is relatively inexpensive, less than 10,000 euros, and it can vastly improve feasibility. It can be used for import purposes when a shipment is being unloaded, or at points farther along in the processing chain. At a workshop we recently held here in Rome, we found that stakeholders had difficulty implementing the recommendation without these instruments. In the harbour at Ravenna, here in Italy, the head of the governmental control authority has informed us that the automatic sampler has made life much easier. It has made the recommendations quite easy to implement.
One thing I see the need for is a worldwide database that includes sampling procedures for GMOs. This type of database could help prevent some of the trade problems we have today. For example, the control authorities at the harbour here in Italy don’t have access to this kind of centralised information source. It is not easy to find out about the sampling procedures other countries may use. There needs to be better communication about sampling at the international level, not only within Europe but also for places like Japan, the US, Australia, and so on.
Dr. Miraglia, thank you for your time and input.
* Co-Extra Deliverable 4.3: Existing sampling plans and needs for the development of novel sampling approaches for Genetically Modified Organisms (GMO) evaluation (not public)