A trove of internal documents from Monsanto, recently unsealed in a lawsuit against the agricultural biotech giant, has revealed the firm’s role in the knotty tale of a paper from the lab of a scientist known for his stance against genetically modified organisms (GMOs).
That paper is “Long term toxicity of a Roundup herbicide and a Roundup-tolerant genetically modified maize,” published in September 2012 in the journal Food and Chemical Toxicology (FCT) and retracted in January 2014. Gilles Seralini, a scientist known for an adversarial stance towards GMOs, was first author. The documents have also spurred the retraction of several pro-GMO articles on Forbes.com written by Henry Miller, a fellow at the Hoover Institution, a conservative think tank at Stanford University. Emails showed that one of the articles, which didn’t discuss Seralini’s paper, was ghostwritten by Monsanto.
The documents, posted last week by the law firm of Baum, Hedlund, Aristei and Goldman, show Monsanto engaged with a network of scientists and other commentators to spread the message that the Seralini paper was bad science and should be retracted. Seralini told Retraction Watch that this proves what he has been saying all along, that Monsanto led a concentrated effort to discredit his science and protect its bottom line. He said:
I deduced from the beginning that it was organized….
But the effort to counteract Seralini’s work was not one-sided. Bruce Chassy, at the time a professor at the University of Illinois, now retired, told us he approached Monsanto to ask if they would act as allies “in the fight against the abuse of science,” in the form of anti-GMO research that he believed to be shoddy. He said that he has been critical of Monsanto’s business practices, but thinks they hire good scientists and are thus a good source of information on GMOs:
If I can use Monsanto to get science a fair hearing, I’ll do it. Since they know more about this than anybody else, they’re the people to go to.
Monsanto tried to ensure that the campaign to retract the 2012 paper wouldn’t be linked back to the company. On Sept. 26, 2012 — a week after the paper came out, Eric Sachs, a Monsanto executive, wrote several colleagues to emphasize:
We have done our part. It is time now for the public sector and especially our network of experts to do theirs…
Monsanto must not be put in the position of providing the critical analysis that leads the editors to retract the paper… There is a difference between defending science and participating in a formal process to retract a publication that challenges the safety of our products.
To achieve this, Monsanto coordinated with the scientists mentioned in the documents on what to say in response to and how to maximize the impact of the response. Based on the documents obtained, that took the form of published articles, including letters to the editor of FCT, at least one of which called for retraction.
A contentious paper
Seralini should be no stranger to our regular readers — controversy has followed the 2012 paper at almost every turn, from the irregular methods Seralini used to promote it to the media (such as asking some journalists to sign a non-disclosure agreement to see it in advance of publication) to initial publication in FCT. Even after it was retracted, it was republished in 2014 — by a journal that didn’t subject it to peer review. The FCT version has been cited 54 times since it was retracted, and 95 in total, according to Clarivate Analytics’ Web of Science, earning it Clarivate’s “highly cited paper” designation.
In a statement issued after the retraction, Seralini accused Monsanto of promoting criticisms of the paper and installing a former employee on the FCT editorial board.
While the documents highlight close ties between Monsanto and at least four people who criticized the study, plenty of scientists criticized the paper’s scientific merits in both the mainstream and academic press. Still, the decision to retract was as contentious as the decision to publish. An FCT investigation found no evidence of fraud, misconduct, or gross error, which are required by Committee on Publication Ethics (COPE) guidelines for retraction; however, FCT cited COPE guidelines in their retraction notice anyway.
As our co-founder Ivan Oransky told CBS News in 2014, after the paper was republished in a different journal:
This whole episode has taken us farther away from knowing the truth …
This is a good example of what happens when people with hardened beliefs manipulate a system for the result they want …
“We are ‘connected’ but did not write the letter”
The documents show that Monsanto was in contact with Chassy, who wrote several pieces attacking Seralini’s paper. On Sept. 25, 2012, Chassy co-authored an article with Henry Miller, the fellow at the conservative think tank, which Miller published in his capacity as a Forbes.com contributor. The article suggested that Seralini had gone beyond “bad science” and into fraud. That article and two other posts from Miller related to GMOs have since been removed from Forbes.com, but are available through the internet archive (one, co-written with Chassy, criticized an earlier 2012 Seralini paper; the other was ghost-written by Monsanto itself in 2015).
Mia Carbonell, senior VP of global communications at Forbes, told Retraction Watch that it has pulled down all of Miller’s articles on its site, because he violated the terms of his contract:
All contributors to Forbes.com sign a contract requiring them to disclose any potential conflicts of interest and only publish content that is their own original writing. When it came to our attention that Mr. Miller violated these terms, we removed all of his posts from Forbes.com and ended our relationship with him.
Both Miller and Chassy said they weren’t involved in the decision to take down the Forbes articles and that they disagreed with the decision. Miller told us that, like Chassy, he has “sought information from [Monsanto] …on scientific issues when I thought they had information of value.” He said he does not have a financial connection to Monsanto. Chassy said that he has never personally accepted money from Monsanto, although his university did [see update].
One day after the Forbes.com article ran, Chassy sent FCT Editor-in-chief A. Wallace “Wally” Hayes a strongly-worded email in which he said the journal had an “ethical obligation” to retract the paper.
Chassy was also one of 25 co-authors of a formal letter to the editor of the FCT that called for the paper’s retraction. (The letter has not been cited.) In September 2012, as the letter was being drafted, a Monsanto staff member asked about the company’s involvement with it and expressed hesitation about whether the company should even disclose its knowledge of the letter to investors. Sachs wrote:
We are “connected” but did not write the letter or encourage anyone to sign it.
As reported by the New York Times, Hayes himself had a “contractual relationship” with Monsanto while he was editor of FCT. Hayes, now an adjunct professor at the University of Massachusetts, told the Times he was not under contract with Monsanto when the paper was retracted and that his relationship played “no role whatsoever” in that decision. However, Hayes was in contact with Monsanto executives soon after the paper came out, offering advice on what they might round up to submit to him as rebuttal.
The docs also reveal ties between Monsanto and Colin Berry, a British toxicologist, who published his own letter in FCT criticizing the paper. On Sept. 20, 2012, the day after FCT published the Seralini paper, Berry emailed David Saltmiras, his contact at Monsanto, to ask:
Have you seen this? I have been asked to comment asap – anything you want to say[?]
Saltmiras replied:
I hope scientist of your caliber and standing may consider penning a joint letter to the Editor focusing on scientific deficiencies and the flawed scientific review process.
A third scientist, Andrew Cockburn, a toxicology consultant who at the time was a visiting professor at UK’s University of Newcastle, was also on the email chain and chimed in:
…if you can provide the ammunition I’d be happy to respond …
Monsanto did not respond to Retraction Watch’s request for comment.
Sachs, the Monsanto executive, realized the risks of attacking the decision to publish the paper and call for retraction. On Sept. 26, as he insisted it should not be Monsanto to swing the axe, he wrote:
We should not provide ammunition for Seralini, [genetic modification] critics and the media to charge that Monsanto used its might to get this paper retracted.
Update 8/11/2017 UTC 14:30:
Bruce Chassy told us that he had never personally received money from Monsanto; however, several readers have pointed out that he has in fact accepted thousands of dollars from Monsanto, routed through the University of Illinois Foundation and nominally slated for “outreach” to avoid scrutiny. A Chicago Public Radio investigation from March 2016 showed:
Monsanto sent more than $140,000 in “biotech research and outreach” payments through the University of Illinois Foundation between 2006 and June 2012, the month Chassy retired. Further documents confirm that at least $57,000 was routed to Chassy over a period of 23 months.
And a New York Times article from September 2015 said that in 2011 Monsanto gave Chassy a grant for an undisclosed amount “to support ‘biotechnology outreach and education activities.’”
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.
Thank you Andrew for this balanced post (I would only put more emphasis on the fact that the retraction notice said the “results presented (while not incorrect) are inconclusive,” making the decision to retraction questionable).
Given the buckets of uploaded internal MON emails to plow through, I appreciated (and RW readers might too) Bloomberg News publishing this analysis of another campaign by MON, to counter IARC’s glyphosate carcinogen listing:
‘Monsanto was its own ghostwriter for some safety reviews.’
http://newsletters.environmentalhealthnews.org/t/280232/177499/210909/0/
Academic papers vindicating its Roundup herbicide were written with the help of its employees. Bloomberg Businessweek.
https://www.youtube.com/watch?v=OWQON4FzQo4
It’s annoying to see so much discussion of the interests that motivate calls for the work to remain published or retracted, seemingly dwarfing discussion of scientific merits.
I have only skimmed the goings-on (including prior coverage by RW), but AFAICT much rests on this phrase in the retraction notice:
“Given the known high incidence of tumors in the Sprague–Dawley rat, normal variability cannot be excluded as the cause of the higher mortality and incidence observed in the treated groups.”
If that is indeed the crux of the matter, why isn’t this a purely mathematical matter, based on statistical assessment of the number of rats in each experimental group, the chance of an effect in each control rat, and the chance of the same effect in each treated rat?
Is there something I’m just not getting?
That is the big red herring. The study was designed as a toxicology analysis and met those requirements. The tumours made the headlines and the criticism focused on that. But that is a deflection. I have seen numerous toxicology scientists support the standards of the study for its intended purpose. Unfortunately the toxicity findings have been overshadowed
There are no statistical differences in that study at all. In any of the analysis.
And the conclusion was basically “more research needed.”
With such small samples, statistical analysis isn’t the big deal most scientists think it is. Toxicology textbooks (including at least one edition that Hayes is editor of) emphasize that *patterns* of findings are more important when the statistical power of preiminary tox-screening studies is so low. Remember: such studies are only have a power of 0.8, assuming an effect-size of 1.0, and a p <0.05, and many potentially important results would be ignored if toxicologists slavishly kept to normal statistical standards when evaluating such a study (paraphrased from Haye's own book).
In the case of Seralini's study, while the sample size was arguably far too small for a full-life study, that particular breed of rat has VERY FEW cancers of any type in the first few months of life, as the historical record of over one thousand SD rats (1,284 to be precise) in control studies shows (see table 3): http://journals.sagepub.com/doi/pdf/10.1080/01926230490440871
When 2 rats need to be euthanized for tumors in an experimental group of 10 within the first few months of a study, while the historical record on 1,284 shows that only a handful die from tumors, and none from that specific kind of tumor, in the first 50 weeks, that is of toxicological interest, even if the statistical analysis does not meet standards of normal scientific research.
I have a PhD in mathematical biology. I am from Russia and I was one of the many scientists who wrote criticism on the Seralini article. You can read my detailed comment here: https://linkinghub.elsevier.com/retrieve/pii/S0278-6915(12)00784-3
The Seralini study had _no statistical analysis_ at all. When I performed the statistical analysis based on the data provided it showed that _there are no significant differences between the experimental and control groups_. That is something anyone with a background in statistics can simply check.
I have no affiliation with Monsanto and the decision to write on this subject was my own. Even if Monsanto had something to do with the retraction, it does not change the fact that is is a poorly designed study, that misleads it’s readers by presenting null-results as findings. It should have been retracted. Actually: it shouldn’t have been published.
Notably, the problem of absent/improper statistical analysis is widespread among “anti-GM” papers. I have written a review on this issue here: http://www.tandfonline.com/doi/abs/10.3109/07388551.2015.1130684
I think this comes from the issue John Ioannidis has been talking about for a long time: http://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.0020124
It’s very difficult to fight off bad science these days.
From the retracted paper:
“Among phenolic acids, the only consistent and significant (p < 0.01) results concerned ferulic acid that was decreased in both GM and GM + R diets by 16–30% in comparison to the control diet (889 ± 107, 735 ± 89 respectively vs control 1057 ± 127 mg/kg) and caffeic acid by 21–53% (17.5 ± 2.1, 10.3 ± 1.3 vs control 22.1 ± 2.6 mg/kg)."
so Alexander Panchin's claim that "The Seralini study had _no statistical analysis_ at all" seems incorrect.
In my opinion the statistics presented are poor, I see no survival analysis statistics and more importantly, no discussion of effect sizes of biological importance and what, if any, power analyses were performed to assess how many observations would be needed to have high power for detecting the effect sizes of biological importance.
In the absence of a disciplined a-priori power analysis, negative findings can not be interpreted as demonstrations of no effect. Absence of evidence is not evidence of absence, if the scientific test was not designed to readily detect magnitudes of effects of importance. Negative findings in this case merely mean that the true state of the outcome remains uncertain.
From the conclusions:
"The results of the study presented here clearly demonstrate that lower levels of complete agricultural glyphosate herbicide formulations, at concentrations well below officially set safety limits, induce severe hormone-dependent mammary, hepatic and kidney disturbances. Similarly, disruption of biosynthetic pathways that may result from overexpression of the EPSPS transgene in the GM NK603 maize can give rise to comparable pathologies that may be linked to abnormal or unbalanced phenolic acids metabolites, or related compounds. Other mutagenic and metabolic effects of the edible GMO cannot be excluded."
The results of this exploratory study do not clearly demonstrate anything, in the absence of disciplined a-priori specified hypotheses, and statistical power analyses to ensure adequate sample sizes to detect effects of interest. This study is at best an initial foray into this issue of glyphosate safety. This data could be used to specify conditions and hypotheses for further studies that could have a chance of demonstrating Roundup toxic effects.
Sadly this paper is just another example of the poor ability of business-oriented journals to thoroughly vet submitted articles for adequate scientific content. Publishing fees trump scientific content. Perhaps if journals had to pay huge fees whenever dodgy papers are revealed and/or retracted, some motivation to present content of actual scientific merit might ensue.
Yes, I forgot to clarify, that I mean the part of the actual “GMOs are harmful” claim. The link between GMOs and cancer stated in the article was not supported by any statistical tests.
The conclusion is obviously far more strongly written than the evidence supports, but preliminary toxicological screening studies are not meant to support or reject hypothesis, per se, but to give toxicologists insight into any patterns of data that might emerge to flag the need for further studies.
That 2 male rats needed to be euthanized at an early age, due to tumors, in a breed of rat that historically has very few tumors at such an early age (see table 3: http://journals.sagepub.com/doi/pdf/10.1080/01926230490440871), should be enough to flag the topic as needing further study or it would in a field where “substantial equivalence” didn’t “prove” that no research such research is ever really needed in the first place.
But preliminary toxicological screening studies aren’t “all about teh statistics,” but about allowing experienced toxicologists to look for suggestive patterns in the data to raise concerns that further study is needed, where samples are large enough that statistical analysis can be used in the normal way.
It should be noted that Seralini himself is an anti-GMO activist, and his apparently poorly designed and conducted study supported his prior views and won extensive support and praise from many people who are opposed to GMOs. It’s hard to say that his position is scientifically superior to Monsanto, and many scientists with little or no connection to Monsanto also criticized the Seralini study. It’s predictable, though, that this would be framed as an evil corporation conspiring against a valid scientific study. Thus, somewhat ironically, scientific criticisms of the Seralini study can end up being taken as evidence that the Seralini study results are correct! It seems like the most reasonable thing to do would be for some neutral party or parties to attempt to replicate the Seralini study with improved methods.
I’m pretty sure that that was the original hope of Seralini and his anti-GMO sponsors.
Here’s an interest set of points from Belgium (unlike the EFSA, the Belgium response included minority versions of evaluation of Seralini’s paper), as summarized by the anti-GMO website, GMOWatch:
http://www.gmwatch.org/en/bills-test/14615-belgian-opinion-on-seralini-study-undermines-efsa-view
[…]
2. GMWatch comments on the Belgian Biosafety Advisory Council opinion on the Seralini et al (2012) study on GM maize and Roundup
4 March 2013
The Belgian opinion can be downloaded here:
http://www.efsa.europa.eu/en/efsajournal/pub/2986.htm
These were the 11 experts consulted by the Belgian Biosafety Advisory Council:
Prof. Adelin Albert (Universite de Liege), Prof. Dominique Cassart (Universite de Liege), Prof. Corinne Charlier (Universite de Liege), Prof. Dr. Dirk De Bacquer (Universiteit Gent), Dr. Bart De Ketelaere (Katholieke Universiteit Leuven), Prof. Joris Delanghe (Universiteit Gent), Prof. Philippe Delvenne (Universite de Liege), Prof. Frederic Famir (Universite de Liege), Prof. Pascal Gustin (Universite de Liege), Dr. Dominique Lison (Universite catholique de Louvain), Dr. Ir. Viviane Planchon (Centre wallon de Recherches agronomiques, Gembloux).
Study design
3 experts endorsed the long duration of the study. 1 expert noted its flaws, but noted that those were common also in studies generally endorsed by regulators.
Only 5 of the 11 experts thought the choice of rat was wrong (so 6 endorsed the use of the Sprague-Dawley rat).
8 of the 11 experts did not criticise the size of the control groups.
Only 1 of 11 experts criticised the number of experimental groups.
Endpoints
Only 1 of 11 experts criticised the endpoint measurements.
Anatomopathological observations
3 of 11 experts thought that the observed tumours should have been characterised. So 8 of 11 did not.
Biochemical parameters (kidney, liver)
Were criticised by only 2 of the 11 experts.
Statistics
Not enough information is given to know how many scientists gave each view, but the statistics aspect of the paper was the most heavily criticised. It is clear, however, that the scientists’ views were very different: there is no sign of consensus on each criticism. Our own experience too is that every commentator on the statistical aspects of any paper has a different view.
The most agreed upon point (4 experts) was that there was no sign in the paper of a statistical analysis of the mortality or tumour endpoints. This is true, but it could be seen as an observation rather than a criticism. Seralini’s team didn’t do a statistical analysis on these endpoints because according to commonly used protocols, such an analysis would need much larger groups of animals. Criticising the absence of statistical analysis for these endpoints would be equivalent to criticising an apple for not being an orange.
Specific conclusions
“The experimental design used in this study allows estimation of the effect of water contamination and of the effect of GMO diet, but not the cumulative effect of both combined, in male and female rats.”
This is a general endorsement of the findings for both the GM effect and the glyphosate effect. Only the combined effect, according to the Belgian experts, is too weak to stand.
“The study provides some indications that GMO and Roundup based diets potentially might have deleterious effects on health, at least in rats. A major result of the paper is that the (potential) occurrence of problems takes time well above the usual duration used for this type of experiences, which strongly indicates that future experimentations should consider longer terms effects than what is usually done.”
This is a general endorsement of a real effect. However, the experts note that replication and extension of Seralini’s experiment is needed, a view echoed in the following conclusion:
“Despite the many methodological shortcomings, it can nevertheless be stated that the results of Seralini et al could rise to further, larger and independent research on the health long term effects of genetically modified food.”
The experts offer the following revealing conclusion – and one that we largely support:
“It seems reasonable to assume that the publication of Prof. Seralini, without providing definitive conclusions as to carcinogenicity in rats and even less about the underlying mechanisms, provides a reasonable and sufficient doubt to promote research on the impact of GMOs and pesticides associated with this type of culture, on the fauna and flora as well as mammals exposed. Rather than rejecting these results, should we not, according to the scientific approach, encourage new experiments to verify the reproducibility of the results by correcting any shortcomings of the current publication. All this calls for extreme caution and to discuss these issues with great care.”
The experts split into two groups issuing majority and minority opinions. The minority opinion asks for the same critical standards as were applied to the Seralini study to be applied to the Monsanto dossier on the same maize:
“Considering the uncertainties on long term effects of GM maize NK603 on health, we ask for a reassessment of the advice of the BAC on the initial dossiers of the maize NK603, regarding effects on human and animal health, using the same critical analysis that was applied by the BAC’s experts to the Seralini et al. study.” – Jean-Claude Gregoire, Damien Winandy, Lucette Ffandroy, and Philippe Baret
We couldn’t say it better ourselves.
What is fascinating about these debates is how scientists have as much difficulty as everyone else from extracting ideological biases and narrow-casting on their particular skill set when approaching scientific questions.
Statisticians don’t see to extract their gaze from p-values to consider the possibility that an effect in an underpowered study which doesn’t reach statistical significance might very well reach significance in a properly powered study. That a study might throw up unexpected effects and hence be underpowered for the trend that does appear to be emerging.
Pro-GM scientists can never get beyond thinking up reasons for why a study that casts a shadow over a particular GM species could be wrong, to thinking for reasons why it might be right. And anti-GM scientists have the opposite problem.
My own view is that the Round-Up ready range is created by adding a transgene with a very powerful promoter and enhancer elements into a major metabolic pathway. That is, the a major step in the aromatic acid biosynthesis pathway has been turned to always “on”. It is certainly possible that there is so much redundancy in gene expression regulation that this has no effect, but it is by no means certain. There is a smaller risk that this might result in a cancer risk on some very defined experiment conditions, but it is not a zero risk. And these issues simply never get properly investigated
Here is a Seralini paper that claims the transformed maize is not metabolically identical – which is not an inevitable result from the crude alteration in the aromatic acid biosythesis pathway but would in no way be surprising
https://www.ncbi.nlm.nih.gov/pubmed/27991589
And here is a critique from a strong GM advocate scientist
https://sciblogs.co.nz/code-for-life/2016/12/31/gm-corn-really-different-non-gm-corn/
The caveats they might could be true, but the GM advocates seem to think because they could be true they are true. When rather the onus should fall on the GM industry to demonstrate conclusively that perturbations from their interventions are insignificant
“And here is a critique from a strong GM advocate scientist”
Not being rude, aside from the straw man that tries to pin a label on me (!) your reply leaves me wondering if you’ve read my article. (But thanks for citing it – I appreciate that.) Firstly, my article takes *other people’s* criticism of the paper (not mine), with my part being to add introductory material so that these might make more sense to those outside of science. (I add two smaller critiques of my own towards the very end – claim to be a first and the regards their title, neither of which are about the scientific content per se.) For another I’m not an advocate (of either “position”), but an independent voice. There are people who are strong advocates, but my position is quite a bit more neutral and nuanced.
“The caveats they might could be true, but the GM advocates seem to think because they could be true they are true.”
This is not what those critiquing where doing — they were pointing out were problems with the Seralini group’s paper. It’s standard practice in science to critique papers, especially ones with obviously flaws.
“When rather the onus should fall on the GM industry to demonstrate conclusively that perturbations from their interventions are insignificant”
If you’re going to apply this, you really have to do it equitably and equally apply it to all ways of creating new varieties – as it’s not how the new varieties are created that creates risk or not, but the traits of the variety. The whole GM thing is quite a red herring. It’s useful to try as an exercise write without mentioning ‘GM’ or ‘GMO’: what it does it force you to look at the particular plant/animal/etc and their traits – the actual rather than the hypothetical.
Looking at GTF’s (Glyphosate Task Force; industry) report to EFSA evaluating the most recent (’14 to 10’16) endocrine disruption glyphosate literature, among the 90% of possibly relevant published studies that they found and eliminated without evaluating their quality was the 2014 re-published Seralini paper. Table 6 of that report (no link available, I got from an NGO’s document request of EFSA), shows the illogic used to dismiss it:
“Republished study: long- term toxicity of a Roundup herbicide and a Roundup-tolerant genetically modified maize. Environmental Sciences Europe, (2014) Vol. 26, No. 1, pp. 1-17
Article retracted. Not considered reliable any more”
–