Séralini paper: Molecular analysis shows GMO corn differs from non-GMO–Is difference meaningful?


In the US, the Food and Drug Administration … GMO crops are deregulated once nutritional and compositional “substantial equivalence” is demonstrated. The set of parameters … necessary to declare a GMO as substantially equivalent  … focuses on a restricted set of compositional variables, such as the amounts of protein, carbohydrate, vitamins and minerals. …

. . . .

Recent technologies used to ascertain the molecular compositional profile of a system, … collectively referred to as “omics  technologies”, are used extensively in basic and applied science. … the majority of authors of [omics studies of crops] conclude that the statistically significant changes observed between the conventional and the GM varieties are not biologically significant because they fall into the range of variations … between different conventionally-bred varieties, and under different environmental conditions….

. . . .

…[W]e have performed proteomics and metabolomics analyses of NK603 (sprayed or unsprayed with Roundup) and isogenic maize kernels.

. . . .

In this report we present the first multi-omics analysis of GM NK603 maize compared to a near isogenic non-GM counterpart. … Although NK603 had comparable nutritional and compositional profiles when originally accessed by the developer company upon registration of their product, our analysis … shows that NK603 grains, with or without Roundup spraying during cultivation, are not equivalent to isogenic non-transgenic control samples.

Several laboratory studies consisting of 90-day feeding trials in rodents have been conducted to evaluate the safety of GMO crop consumption. These investigations have frequently resulted in statistically significant differences in parameters reflective of disturbances in various organ systems and in particular liver and kidney biochemistry, but with interpretation of their biological significance, especially with respect to health implications, being controversial.

For more background on Seralini’s previous GMO studies and the funding for his research–this study was funded by an anti-GMO organic group, as listed in the paper–read the GLP’s Biotech Gallery profile.Editor’s Note: The GLP has asked independent scientists to review this study. We will update this comments as they come in:

Seralini does not appear to understand the legal concept of “substantial equivalence”

1) There is no information as to whether he looked at compounds on the OECD list. These were chosen for their relevance to detect changes in key metabolic pathways. His list needs to be compared to the OECD one.

2) Once a difference has been found, the second step is missing altogether in this paper. Namely, all statistically significant, biologically relevant changes need to be compared to values in other varieties. If they are in the normal range, there is no issue. The actual differences observed are small, many much smaller than the differences found from one field to the next for the same crop. Statistically significant differences are not biologically significant differences.

3) At the end of the day, the OECD list covers about 60 metabolites that make up 95% of the grain. The remaining 5% contains many tens of thousand metabolites, where the “dose makes the poison” adage kicks in.

4)  Differences in proteomes are for the most part irrelevant to food safety assessment. The reason for this is both simple and complex. Stated simply, large differences in proteomes are often seen in specimens with essentially identical compositions. The underlying complexity is that the concentrations of metabolite(s) produced by a protein or a set of proteins in a pathway reflect the interaction of a panopoly of variables such as enzyme concentration, substrate concentrations, concentrations of regulators, metabolic fluxes within the organism, and environmental regulation. Maybe a simple way to put it is that under the conditions at which a cell operates a great deal more regulates the concentration of a particular metabolite than the concentration of the enzyme that catalyzes its synthesis. I can’t tell you how many papers there are that report that enzyme X was elevated 2-fold or 5-fold or 100-fold but no change was observed in the product of its metabolic product. The rate limiting step in the production of a metabolite is not necessarily dependent on how much of the enzyme that produces it is present. Changes of a few fold are often inconsequential, and changes of a few percent sometimes evoke larger changes than one might expect. At the end of the day the proteome doesn’t tell us anything about safety. Proteomics can be a valuable research tool in the hands of an investigator with an hypothesis, but as applied here its simply another data dredging tool looking for a statistically significant change that means nothing.

5) What really counts from and safety and nutrition perspective is the metabolome. That’s the part we eat isn’t it? All of the changes in minor metabolites that were reported are small compared to the magnitude of changes in minor metabolites that is often observed in the same crop plant or its seeds. They are also meaningless changes; a few-fold change in polyamine concentration has no biological significance and if this experiment was done 10 times with NK603 the polyamine concentration would probably be all over the place.

6) There are some more fundamental problems with omics as an assessment tool which are covered in my 2010 paper. In particular, Lay et al published a compendium of troubles that face -omics as a science tool. These include the more obvious data dredging and statistics arguments to some pretty interesting math that indicates in some cases omics is more likely to give a wrong answer than a correct one. Think of it this way, the science of analytical chemistry has spent the last few hundred years developing methods for measuring one compound with accuracy, precision, reproducibility, sensitivity, etc. Omics is the emerging science of measuring everything poorly. When used correctly and in competent hands, omic analysis can be a powerful research tool. What it can’t do is tell you if two varieties of corn are equally safe to eat. It can, however, tell you that they differ in some way, and to an investigator in lab that may (or may not) be important to know, however, since there will no doubt be 100s if not 1000s of differences observed, good luck figuring out what they all mean. Omic analysis may eventually provide high quality useful data that tells us something about safety, but we are not there yet. Which of course means that the paper in question should be disregarded as artifact. Ir was in fact surprising to see that they did not observe more and larger differences.

The GLP aggregated and excerpted this blog/article to reflect the diversity of news, opinion and analysis. Read full, original post: An integrated multi-omics analysis of the NK603 Roundup-tolerant GM maize reveals metabolism disturbances caused by the transformation process

  • Farmer with a Dell

    I can hardly wait for the scientifically illiterate mental midget anti-GMO trolls to crank up their absurd rhetoric on this one. Jumpin’ Jeebus, each and every one of those pathetic assclowns will present himself/herself as a goddam authority, not only on “omics”, but on every friggin’ molecule dredged up in this asinine witch hunt. Seralini now supplies us with a beautiful example of the old cliche: ‘If you can’t dazzle them with brilliance, baffle ’em with bullshit’.

  • Robert Howd

    Interesting paper, but why would one call differences in chemical profiles “metabolic disturbances” without any information as to whether these differences have any functional consequences? The authors don’t note, for example, the relative yields and health of the GMO and non-GMO plantings. If the GMO corn was unhealthy, shouldn’t this be documentable? They note that studies of this type show wide variations in metabolic profiles, so how similar are the differences shown here to the differences found in other studies? Obviously the wide-spread adoption of various GMO cultivars in the U.S. and other countries represents farmers’ conclusions that the GMO plants are healthy and productive. Also, the dire predictions of the authors that the chemical differences they have observed means that “nutritional quality of GM feed might be hampered by metabolic imbalances” is belied by the observations of farmers feeding millions of animals the GMO products over the last few decades, as reported by Van Eenennaam and Young in J. Animal Science 92:4255-4278, 2014.

    • Kevin Folta

      That’s my point too Robert. They are not “disturbances” they are differences, and differences that are likely within the range of variation you’d see between two adjacent plots in a field. Good for them to do such tests– this is a great step. They just showed that the products are essentially the same.

  • Last year there was an article where they computationaly predicted increased formaldehyde levels in gly resitant crops. http://www.scirp.org/journal/PaperInformation.aspx?PaperID=57871#.Vbfui3hbv6Q I wonder if Seralini confirmed this finding experimentally.

    • Robert Howd

      No, that wasn’t evaluated.

  • Wackes Seppi

    « At the end of the day, the OECD list covers about 60 metabolites that make up 95% of the grain. The remaining 5% contains many tens of thousand metabolites, where the “dose makes the poison” adage kicks in. »

    You are bloody wrong !

    The remaining 5% contains the endocrine disruptors, where the “dose makes the poison” adage DOES NOT kick in !

    Of course, I am kidding.

    I am quite surprised that the only financial acknowledgement goes to the Sustainable Food Alliance.

    By the way, according to this recent paper :

    « Conflicts of Interest in GM Bt Crop Efficacy and Durability Studies », Thomas Guillemaud, Eric Lombaert, Denis Bourguet, http://dx.doi.org/10.1371/journal.pone.0167777

    this is to be as of itself a conflict of interest (of course the authors declare no conflict…).

  • Kevin Folta

    My favorite part of the paper is that they did NOT detect glyphosate on plants sprayed with glyphosate. However, activists claim to detect it in food.

    The rest of this paper confirms well that the products are essentially the same. The differences observed are not much more than you’d expect from small environmental variations in plant biology. I would have liked to have seen a comparison within samples from the control group (the isoline). I have a funny feeling you’d see variation there too. Small differences in moisture, etc could account for the differences.

    On the other hand there could be small collateral changes induced by a transgene. No surprise there. The question is, is there any reason to believe the changes observed in metabolites are problematic? No. Not at all. Other plants make the same polyamine compounds in mountains relative to corn.

    The title and discussion were completely inappropriate for a scientific journal and should have been revised. But obviously soft reviewers and editor that let it slide.

    • mem_somerville

      Wait, they claimed it was in there in their last paper–that awful mouse chow study for which they wouldn’t release their data.


      In fact, they said:

      The main pesticide detected was Roundup, with residues of glyphosate and AMPA in 9 of the 13 diets

      Hmm. Curiouser.

      • mem_somerville

        Wait, the top protein change in their list is: http://www.uniprot.org/uniprot/W7LNM5

        That’s a rot fungus tubulin. They had infected corn? That sort of means all their other claims are…umm….infected, perhaps?

    • Roy Williams

      It is really, really disappointing that Nature published this. Hopefully people with appropriate credentials will write to the editor of Nature.

      • Kevin Folta

        Hi Roy, it was not Nature. It was Scientific Reports, in the Nature Publishing Group. It is not a bad journal, some lousy stuff here and there and improving. This work clearly snuck through. The title alone oversteps the data. But ultimately it shows that the lines are really similar, likely within biological variation, and it goes away as a major snooze. Glad that they are doing this level of analysis. They just don’t know how to properly design a crop experiment.

  • RobertWager

    I am curious about DKC 2678 vs DKC 2675 are described as “nearest isogenic non-transgenic control. Does anyone know if this is accurate or are there DKC 2676 and DKC 2677 that could have/should have been used?

  • Rod Herman

    If I understand the source of the grain samples, there was a single source of each test entry raised at one site. Claiming a difference due to test entry (variety) is like claiming my crop variety yields higher than your crop variety based on a single plot at one location. Considering identified deficiencies in the experimental design from previous publications from some of the authors of this study, it is surprising that the editors did not run this by someone with field experimental-design experience. The design does not appear to allow the reported effects to be attributed to the test entries. Variation in composition and phenotype (like yield) is well known to be influenced by micro-environment, epigentics, etc, Field-plot replication is the only way to tease out whether the crop variety has any impact on the results or if one is measuring noise. Furthermore, isolines are never actually genetically identical to GM lines, so any effect would need to be demonstrated in multiple genetic backgrounds to be convincing. IMO – The current study suggests a hypothesis, but it does not test it.

  • Peter Olins

    Not everything that can be counted, counts; and not everything that counts can be counted.

    I, sadly, do not know enough statistics to be able to conclude whether the differences seen by Mesnage et al. were truly “significant”. However, given the shameful distortions seen in so many previous papers from the Seralini crew, I have the nagging suspicion that there may be many ways to massage constellations of thousands of data points in order to find a desired “effect”.

    Looking at the larger picture, there are literally hundreds of pesticides and formulations applied to edible crops, and it seems highly likely to me that, given very sensitive detection techniques, many differences will be seen under different conditions. But so what? Do we really need mass-spectroscopy on every bag of Cheetos that we buy?

    Unfortunately, this paper lends support to the recent recommendation from the National Academies — misguided, in my opinion — that omics analysis should be used as a way to gauge the safety of genetically engineered crops.

    (Finally, I note that the group applied 26 tonnes of “liquid dairy manure” per hectare of crop: it would have been interesting to see the results without such a large application of BS.)

  • Pogo333

    One of my great frustrations with Seralini’s field-to-lab papers is the lack of detail on the field side. Okay, they treated RR corn with glyphosate (at least they shared the rate in this paper – 3 liters per hectare of Weathermax), but when and how? Was it treated when the corn was small? At tasseling? When ears were filling? Timing of application greatly influences residue amounts and locations. Was it applied by air? High clearance sprayer? Any adjuvant? Any rainfall between application and sampling? How long after application were the samples collected? Any other chemicals applied? How were weeds managed in the non-RR fields, and were they a problem there (weed competition can affect crop physiology)?

    This paper at least provided some detail on soils, crop spacing, and fertilization, but they still seem disinterested in other critical details in the field that shape what they find in the lab – at least disinterested enough that they fail to share it with readers. And the harsh reality is that the lab assays are the least significant part of the project. In the lab they are only evaluating what has already happened in the field using what are basically cookbook and automated methods. What occurred in the field is the key, and they give this critical part of the story very short shrift in their methods.

    • Farmer with a Dell

      More than likely the Sustainable Food Alliance made clear it wasn’t paying Seralini to fart around getting the agronomy right, nor were they concerned about confounding variables of any sort. Nope, SFA was buying results with which to smear GE technology. The bang for the SFA buck comes from hyperbole in the title of the published paper, from the author’s hypotheses and conclusions that can only be politely described as a s-t-r-e-t-c-h, and from the click bait all that provides. Validity of the science isn’t even an afterthought. No one at SFA even understands science, much less respects it.

    • Alokin

      Plant stress due to hypoxic soils or saline soils, among other factors can affect putrescine levels. Authors mention stratified, “imperfectly drained soil.” Absent any treatment effect, it would not be unusual to see differences in putrescine concentration within such a field trial.

      • RobertWager

        Can you please give a reference. Thanks

        • Alokin

          Let Google be your friend: http://bit.ly/2ias4z8

          This is basic plant physiology, so I didn’t provide a specific reference. There are lots of them.

          • RobertWager

            Thank you. A great list of references. Cheers

    • Rod Herman

      The paper does disclose that results are from single samples which is sufficient information to invalidate the attribution of any findings to trangenesis or anything else. One cannot determine if a “signal” is real without measuring the background “noise”. That is what replication of samples (replicated analyses from replicated plots) allows one to do. It appears that the “glitter” of analytical technology distracted reviewers and editors from detecting this basic scientific flaw.

      • I am not suprised by this. It is not the first paper of this team and every single one so far was designed in such a way that it would record and amplify any noise and then claim that there is statistically significant difference. What suprises me is that they always find reviewers willing to accept it. I wish I could also be so lucky.

        • Rod Herman

          Be glad that “you are not so lucky”. This is not the kind of legacy you want to leave!

        • Mark Mattingly

          They design something from the start that will generate inconclusive results. They do the study and make all kinds of claims. They are successful, because the work doesn’t rule out whatever the hypothesis.

          • given their huge experimental budget I even suspect that they report only on part of their datasets. Eg in this work they could have analyzed several control field plots and reported only those with lower cadaverine and potrescine content, just because the PR potential of these chemicals.

  • Jason

    At some point, somebody needs to tell Seralini that the horse is dead and there’s no need to continue beating it.

  • My hovercraft is full of eels

    Seralini. I just have to hear his name and I start laughing at the bad science that is sure to follow.

    • Realiτy022

      Ahemm….E-acute αsshole.

      You probably think Séralini is a type of pasta.

  • Grant Jacobs

    I’ve added a bit of round-up of what people have said about this paper on Sciblogs. I didn’t touch on the glyphosate aspect – sorry Kevin, just ran out of time…! I would like to have mentioned that you mentioned that the differences were basically in the range you’d expect. Maybe if I get time to edit it.


    • Grant Jacobs

      I’ve quickly updated it now.

  • The FDA have defined substantive equivalence as, ‘the materials, ingredients, design, composition, heating source, or other features of..’ http://www.fda.gov/TobaccoProducts/Labeling/RulesRegulationsGuidance/ucm262073.htm#910_A_3_A