I am getting the "message" that we need to plan a quantitative PCR collaborative. If we focus on corn samples, can we assume that other grains will respond similarly? Do we need to evaluate processed product as well?

Whether to use processed food fractions depends on the objectives of the study. If the point is to determine if some number of labs running the exact same procedure can come up with the same answer, then it may not be necessary to have processed food fractions. If the study is looking further to validate the use of a method with certain food fractions then, just like any other method validation, it is necessary to validate the method with the samples that the method is intended to be used with. Heinz Schimmel has stated repeatedly that the amount of processing of corn effects the result he gets from PCR. I can't recall if he has determined if that is due to the quality or size of the DNA, matrix, extractibility or what. I do recall that he believes it is related to particle size.

What samples to measure? I suggest you start at the end. Basically, GM-material can end up in a wide range of foods, but there are some mainstreams. Analyse those and take out the typical ones representing product categories or specific processing methods (drying etc). Then, take a step back towards the source and select some of the intermediary products. To keep things practical you should not go beyond 10 samples. On the other hand, exploring typical sample preparation related problems at an early stage (in my mind) is pure wisdom.

I believe it would be wisest to restrict ourselves to ingredients at the beginning, and separately evaluate corn/soy/canola (if we decide canola should be included). Since lab variation is such a problem, we should first show that we can get consistent results on the less-processed materials, before we try finished products. In my lab, we too have had trouble recovering DNA from highly processed material. So I suggest we restrict the initial attempts to ground whole grains, soy protein, corn grits, lecithin, etc. If the method works, we can extend it (perhaps with some modifications) to more processed materials for a second trial.

I think it would be best to begin with a PCR study using grain. This would provide us a good initial measure of the consistency of results between labs. As you move to more processed ingredients you have to deal with difficult problems related to the presence of limited good quality DNA remaining in the sample after processing and extraction procedures to isolate sufficient good quality DNA. I think it is important to first establish the reliability of a PCR procedure on a sample like grain before we move to more difficult samples.

The first samples should be whole grain, ground ready for testing. The next step could be whole grain samples to test sample prep procedures and any back to front contamination problems. Later, ingredients such as unmodified corn starch, soy isolate etc. could be tested. I think the first stage should be as simple as possible to allow the "beginning" labs to catch up to the "advanced" labs.

In addition to opinions 4 to 6 I think the GMO percentage of the samples should be from 0,1% to 5% for quantitative. If we are not only talking about proficiency testing but method validation there are other aspects to be defined or adressed: Number of replicates, controls, specificity of the assay, nature of the calibration material to be used ...

In response to Ralph's posting: I trust he is not suggesting that "advanced" labs that have technical capability, or labs which are implementing a techncial advance, should not offer those advanced services until/unless "beginning" labs can offer the same services. I am pretty sure that clients want to have the best available technology applied to their samples.

Andreas, I would also very much like to see samples at 0.01% GMO or even less for this first round. We don't have concrete data on what levels will still give a positive result for the 35S screen but we believe 0.01 still gives a strong positive in my lab, based on the results of some spiked samples.

I get the impression from this discussion that each laboratory would run their own method. How is this different from previous studies? If all the labs don't get the same result what conclusion do you draw? Is the idea to validate the samples, the method(s), or the laboratories? If you want to know about the variability of a method between labs, every lab has to run the same procedure. If you want to validate the detection limit of a method, you first determine during method development what the detection limit is, and you use the study to 'validate' that the method can meet the claims. A collaborative study is generally not used to develop the method, but to validate the performance once it has been defined. Similarly, a collaborative study is not usually used to determine if a method is appropriate for use with a particular type of sample, rather it is an opportunity to evaluate the performance against stated claims. During development the method would already have been shown to perform with the particular sample type.

I believe the studies that have already been completed have demonstrated that different labs using different procedures will get different results. It would be useful to have data regarding a specific method (like an official AACC or AOAC method) rather than a generalization about a technology.

Jim, I've been assuming that we all would be running the same procedure - or, at least, the same chemistry. I am referring to situations similar to the PE TaqMan chemistry, which is licensed to other companies for use with their detection systems (e.g. Bio-Rad). I realize that this would mean that not all of the current participants would be running the actual study, if they have committed within their company to a chemistry which is different from the one we choose for the collab.

Please read also my comment I made concerning the former topic on criteria for quantitative methods vs qualitative methods. I fully agree with the comment made by Jim Stave. Processed samples are needed to fully validate a method. As stated in the above mentioned comment we observed discrepancies between RR soybean samples which contained the same amount of GMO but which were processed in a different way. Since the GMO's end up in a large variety of products and since the production processes of commodities can vary enourmously I would recommend to carry out the validation studies using a limited number of well defined materials and to focus on raw products and basic commodities. This kind of study can be carried out by a limited number of pilot laboratories. There are some guidelines how validations should be carried out. These guidelines have been developed for chemical parameters where one determines the absolute concentration of an analyte (identical with the parameter to be determined) in a sample at the time it is analysed. However the question does not fit to the GMO problem and therefore we think that specific guidelines for the validation of GMO measurements need to be developed. In the field of GMO measurements the parameter to be determined (i.e. weight % GMO) is only measured indirectly via proteins or DNA as the analyte. Therefore compared to the chemical validations one needs to understand in addition what is the correlation between % GMO and the analyte concentration and what happens to the analyte if the material is processed or analysed. Possible deviations in the degree of DNA degradation influence the quantitation even if normalised via a housekeeping gene. The particle size distribution and the composition of different seed parts, e.g. the oily embryo versus the endosperm, may have an influence on the extraction efficiency and probably on the quantitative result due to the fact that the genetic composition of the embryo and endosperm in maize is different. Another question to be clarified for DNA measurments (for protein it is obvious) is, whether the concentration of DNA in non-GMO or GMO is constant or subject to variation which would have a direct impact on the GMO quantitation via DNA. Therefore one needs to run method validations in a much broader sense and not only with a limited number of samples in order to get a realistic estimation of the uncertainties related to the measurement process. If one decides to leave the choice of the method to be used for GMO detection up to the laboratories as long as they use a validated (the question is how far ?) method then ring trials should be organised with laboratories using these different methods in order to come up with a realistic estimate of the uncertainties. Since there have been concerns that the scheme might become too complicated especially for newcomers I would suggest I stepwise approach. One could start with DNA preparations. However these would only cover the final determination step and would be to some extend a help to identify inhibitions. Purified DNA would not be suitable to determine extraction efficiency. Next step could be to run ring trials on raw materials and next processed materials. Following this approach I only see the difficulty that you may discover problems at a very late stage because a number of crucial points like unequal DNA degradation will only be discovered with the last step. A last word concerning the quality of materials to be used especially if one starts to talk about 0.01 % materials. The materials need to be produced under fully controlled conditions and to be of highest quality because the whole study will be based on these materials. Making a 0.01 % material (e.g. maize or soybean powder) goes far beyond simple grinding and mixing. It would be a major challenge to get homogeneous material due to natural limitations like the cell size and difficulties in obtaining small enough particles.

Heinz, I am not a biotechnologist and so some of your terminologies are new to me. What do you mean by %GMO, DNA or analyte in a sample? Do you mean the % GMO material in a sample? And if processing degrades GMO or DNA, does it mean the final processed product will be less contaminated by the GMO material, or still have the same level of contamination? Thanks