The NAS report on GM crops and foods ‘disappears’ findings of harm and potential toxicity in animal feeding studies with GM crops and elevates flawed reviews claiming safety, writes Claire Robinson
“GM food safe to eat, say world’s leading scientists”, ran the headline in The Times in the wake of the publication of the US National Academy of Sciences report on GM crops. In almost all the mainstream media the story was the same, from Nature World News trumpeting, “Scientists declare GM food safe”, to The Guardian’s “GM food generally safe for humans”.
In reality, the part of the report that deals with animal feeding studies on GM crops is a subtly treacherous mix. Scattered among some sound statements and useful recommendations are a plethora of strategic omissions, gobsmackingly unscientific assertions, wishful thinking, pulled punches, and outright lies. Below I consider just a few.
The NAS ‘disappears’ ill effects on GM-fed animals
The report’s most outrageous deception is the obliteration of the many findings of harm or risk in animal feeding studies on GM crops.
The report says, “The research that has been conducted in studies with animals and on chemical composition of GE food reveals no differences that would implicate a higher risk to human health from eating GE foods than from eating their non-GE counterparts.” That message was translated by the NAS’s press release as “no substantiated evidence of a difference in risks to human health between current commercially available genetically engineered (GE) crops and conventionally bred crops”.
That’s where the media got the message that GM crops are safe. It wasn’t (just) dumb or lazy reporting. It came straight from the NAS itself.
But both statements are at best misleading and at worst lies, as anyone knows who has seen any of the long list of animal feeding studies showing risks and harms from GM crops. Ill effects in GM-fed animals include liver and kidney damage, changes in blood biochemistry, and immune responses.
Some might argue that animal studies are not necessarily applicable to humans and thus the NAS’s careful wording of risks to “human health” is defensible. But experiments on animals, especially rodents and pigs, are mandated by regulators worldwide to test and assess the potential human health impacts of pesticides and other chemicals, as well as (in some countries) GM crops. As a society, we’ve agreed on this system, and so we must take seriously the findings of animal studies.
The GMO industry and its allies are well aware of this and fight hard to try to persuade regulators not to require animal feeding studies with GM foods and their associated pesticides – and shoot down those that are carried out and that find problems.
And for whatever reason, the NAS also seems to have felt it necessary to ‘neutralize’ the animal feeding studies that have shown problems with GMOs.
How do the NAS do that? By avoiding directly addressing the findings of harm or signs of possible toxicity in the relevant studies. As far as the NAS is concerned, these studies may as well not exist. Instead they elevate to a position of authority two misleading reviews, written by conflicted-out authors, which claim to find no evidence of harm in GM-fed animals.
Misleading review 1: Van Eenennaam and Young
One of these reviews simply failed to look for any meaningful health effects (Van Eenennaam and Young, 2014). The Van Eenennaam and Young review claims to consider 29 years of livestock productivity field data in the US from before and after the introduction of GMOs, representing more than 100 billion farm animals. These data, the authors said, do not show that livestock health declined after the introduction of GM feed, and therefore show that GMOs cause no health problems in animals.
However, the review tells us almost nothing useful about GMO safety. Many major problems with the review have been identified by the veterinarian Dr Ena Valikov, a vocal critic of GMO science spin and of the Van Eenennaam review in particular.
Nearly 95% of the data are on 49-day-old chickens
Dr Valikov points out that nearly 95% of Van Eenennaam’s data are on 47–49-day-old broiler chickens. Chickens are an irrelevant model for assessing human or even mammalian health risks. And given that a chicken’s natural lifespan is typically 5–7 years, a 49-day-old chicken doesn’t tell us anything about long-term health effects in any animal, even chickens. As Dr Valikov writes, “Even if the study reported convincing health data, it would remain a very short-term study. In other words, it is 19 years’ worth of 49-day-old chickens, which is quite different than 19 years of life-long animal studies.”
No medically relevant data
There are no medically relevant data in the review, so no conclusions can be drawn about health.
Dr Valikov says that for a study drawing conclusions about health, she’d expect to see data on incidence rates of various diseases, laboratory research, post-mortem examinations, and histopathology (microscopic examination of tissues). Such data on livestock health were collected by the FSIS (USDA Food Safety and Inspection Service) between 1998 and 2002. Dr Valikov judges these to be the only medically meaningful data on livestock health. There are no FSIS data from before GMOs were introduced in 1996 or after 2002.
In contrast, the chicken data in the Van Eenennaam review are on the number of days to market, feed efficiency (ratio of feed eaten to weight gained), percentage mortality, and carcass weight.
Dr Valikov explains, “Livestock performance is not a marker of health because the goal of livestock production is minimizing inputs and maximizing production of meat, eggs or milk, regardless of costs to the animal's health or longevity.
“For the rest of us, who aren't slaughtered at 49 days, the goals are completely different and extrapolation of findings in livestock to non-livestock – either pets or humans – is absurdly unreasonable.”
For example, rapid weight gain, which is desirable in livestock and is counted as a positive indicator of health by the Van Eenennaam review authors, is harmful to the health of non-livestock animals and humans.
Condemnation rates not an indicator of health
The number of animals condemned at slaughter as too unhealthy to enter the food supply has decreased over the years since GM feed was introduced. The Van Eenennaam review authors suggest that this shows improving health. But Dr Valikov disagrees: “Speed of slaughter lines and the number of assigned federal meat inspectors, whose numbers have been dwindling, affect condemnation rates. They do not substitute for proper post-mortem analysis with histopathology.”
The experience of Australian scientist Dr Judy Carman seems to confirm that slaughterhouse inspections only pick up the most grossly obvious disease, such as large tumours. Dr Carman’s pig feeding experiment (ignored by the NAS) found that GM-fed pigs had a higher incidence of severe stomach inflammation and heavier uteri than non-GM-fed pigs. But Dr Carman said all the pigs passed the slaughterhouse inspection, though some were clearly unhealthy.
NAS draws unsupported conclusion
The NAS admits the short-term nature of the data in the Van Eenennaam review, stating: “Of course, most livestock are slaughtered at a young age, so that data cannot address the issue of longevity directly.”
But it goes on to draw a completely unsupported conclusion about health: “However, given the general relationship between general health and longevity, the data are useful.”
Yet there are no meaningful data on health in the study, so no conclusions can be drawn about longevity. Far from being useful, these data are useless.
Uncontrolled variables
The data from Van Eenennaam and Young’s “100 billion animals” are uncontrolled for many variables. It is not even known how many of the animals were eating GMOs, in what proportion of their diet, and for how long.
Other uncontrolled variables include escalating antibiotic use in livestock, which can mask health problems. Antibiotics also have the effect of making animals gain weight more rapidly on less feed – that’s why antibiotics used in livestock farming are called growth promoters. This is another reason why the increase in feed to weight gain efficiency observed in livestock since the introduction of GMOs is not a marker of health.
How the NAS’s experts managed to miss the above points, which should be as plain as daylight to a scientific mind, is anyone’s guess. The NAS’s elevation of this absurd study led Dr Valikov to accuse the Academy of “attempting to turn chicken shit into diamonds”.
Misleading review 2: Snell and colleagues
The NAS also relies upon another review, by Snell and colleagues. Snell and colleagues did look at studies that found toxic effects in animals fed GM foods – but then dismissed the findings by sleight-of-hand. Toxic effects included enlarged lymph nodes in mice fed GM glufosinate-tolerant triticale (a hybrid of wheat and rye) over 5 generations and more acute signs of ageing in the liver of mice fed GM soy over 2 years.
Snell and colleagues dismiss these effects on the grounds of certain methodological weaknesses in the studies, such as that they failed to use the non-GM isogenic line (the non-GM parent of the GM crop), grown in the same conditions, as the comparator for the GM crop. The NAS also correctly draws attention to this issue as a general problem in GMO crop feeding studies.
But as is clear from Snell and colleagues’ review, studies that conclude the GM crop tested is safe, and studies that conclude harm or risk, both suffer from this same limitation. Yet in an example of unscientific double standards, Snell and colleagues accept at face value the studies that conclude safety, while rejecting as unreliable the studies that find risk and harm – even though they suffer from the same weaknesses.
As the European Network of Scientists for Social and Environmental Responsibility pointed out, “Based on this asymmetrical, result-triggered approach, the review incorrectly concludes that no health hazards were found in 24 analyzed publications.”
No wonder Marc Lavielle, scientific committee member of France’s High Council of Biotechnology (HCB), told Le Monde that the Snell review was “biased”.
The NAS committee follows Snell and colleagues into exactly the same pattern of result-based double standards. It uses methodological weakness as a reason to dismiss or ignore studies that find harm from GMOs, yet it accepts studies finding safety that suffer from exactly the same weakness.
Statistical significance and biological relevance
Another scientific failure of the Snell review authors – and in some cases the authors of the original studies reviewed – is that they dismiss significant differences in the GM-fed animals as not biologically relevant, without defining what biological relevance actually means, or setting a limit for how serious or large an effect has to be before it is considered biologically relevant.
The NAS, to its credit, sees the problem with this practice and correctly says that researchers should define in advance how large any difference in the GM-fed animals has to be in order to qualify as “biologically relevant”. But at the same time, in order to reach its conclusion that GM crops are no more harmful than non-GM crops, it must have accepted at face value the claims of a lack of biological relevance that have been made about all the adverse effects found in studies. If it were acting in the public interest, the NAS should have dismissed the claims of a lack of biological relevance pending further evidence, and concluded that no conclusion of safety can be drawn.
Once again, the NAS has applied unscientific double standards to give GM crops a clean bill of health.
The NAS wants further study – or maybe not
Again to its credit, the NAS agrees with critics of GMO safety that worrying findings in animal feeding studies should be followed up with further research, though it carefully avoids admitting that any of these findings were of harmful effects, preferring to use the term “equivocal results”:
“Public funding in the United States should be provided for independent followup studies when equivocal results are found in reasonably designed initial or preliminary experimental tests.”
But one has to wonder what researcher would even bother to do an honest investigation of the safety of a GM crop, given that the NAS and a compliant media have effectively wiped out all previous findings of adverse effects with their declarations that no differences have been found in GM-fed animals and that GM foods pose no higher risks than non-GM foods. And given that even if the rats fed the GM crop develop clear signs of organ damage, the results will be termed, at best, “equivocal”.
Just in case any researcher is bull-headed enough to ignore the implicit message here – “Don’t ever find a problem with a GM crop!” – the NAS has another argument ready. It’s one that would bypass anyone who is not familiar with this research field, as it’s slipped into the report almost as a by-the-way. The NAS says that any study measuring multiple outcomes, “whether for GE crops or any other potential toxicant, could require over 1,000 animals to obtain reasonable statistical power”. It reaches this conclusion by applying a statistical correction called the Bonferroni method.
Statistical dodge magics away findings of harm
The Bonferroni method has already been used in a review of animal feeding studies with GMOs to dismiss all adverse effects. It’s titled, revealingly, “Published GMO studies find no evidence of harm when corrected for multiple comparisons”, and argues that these studies provide only “weak evidence of harm that cannot be differentiated from chance”.
The problem with the Bonferroni method, as pointed out by many scientists, is that it increases the incidence of Type II errors, or false negatives. That’s when an important effect exists but is wrongly deemed not to be significant.
The Bonferroni method is a GMO proponent’s dream. We can expect to see it invoked whenever an animal feeding study finds a problem with a GM food – though no such doubts are raised by GMO proponents about the statistical power of studies that claim a GM food is safe, even though they typically use similar numbers of animals. Neither does the NAS allow statistical issues to challenge claims of safety in GMO feeding studies.
To witness the NAS joining in an abuse of statistics to disappear inconvenient findings is a sorry sight.
NAS misrepresents the Séralini study
I’ve come to see how someone treats the Séralini study on GM maize and Roundup herbicide as a barometer of their good faith with regard to GMO safety. In common with all studies, it has strengths and weaknesses. Since it was designed as a chronic toxicity study and not as a carcinogenicity study (as was stated in the introduction to the study on its first publication), a fair analysis should focus primarily on the toxicity findings, even though it was the observation of multiple tumours that attracted much of the publicity.
The toxicity findings were that rats fed GM maize and Roundup, both together and individually, suffered liver and kidney damage and hormonal disruption. While the study also noted an increase in tumours and mortality in the rats fed GMO maize and Roundup, the study did not have enough animals to qualify as a carcinogenicity study, so the tumour and mortality observations need to be followed up in studies with larger numbers of animals.
This study was retracted after a campaign of sustained pressure on the editor of the journal that published it, and following the appointment of a Monsanto-linked scientist to the editorial board. It was later republished by another journal.
How did the journal justify the retraction? By treating the study as a failed carcinogenicity study, rather than what it really was – a chronic toxicity study that fell squarely within the norms for such studies. The editor of the journal stated that the tumour and mortality trends were “inconclusive”. Hundreds of scientists protested that retraction of a study could not be justified on such grounds. But pro-GMO lobbyists and even the European Food Safety Authority (EFSA) kept up the pretence that the study was a failed carcinogenicity study, seemingly as a way to dismiss its other, more solidly based, findings.
It is shocking to see the NAS follow suit. While acknowledging that the study’s authors described the study as a chronic toxicity study and not a carcinogenicity study, the NAS completely ignores the chronic toxicity findings – the liver and kidney damage and the hormonal disruption in most treatment groups. The NAS’s justification? “The committee’s analysis focused on the tumor data because they have received the most attention from the public and news media.”
When I read this, I had to do a double-take to ensure I hadn’t momentarily been transported to a parallel universe. The NAS is supposed to be led by science, not the prevailing media buzz.
There seems to be only one possible explanation for the NAS’s perverse action: it had to find an excuse not to deal with the strongest scientific findings of the study.
Incidentally, a separate study has confirmed through a gene expression analysis that the rats fed the lowest dose of Roundup in the Séralini study did indeed suffer liver and kidney damage. The NAS conveniently omitted it from its discussion of the Séralini study, leading to the conclusion that the Academy, to borrow the words of T.S. Eliot, cannot bear very much reality.
Lobbyist’s dream
The NAS report contains some useful analysis and sound recommendations – which almost certainly will not be acted upon. But on the vexed question of GMO safety, it is profoundly misleading and unscientific.
The NAS’s press release announcing the report hits all the current GMO lobby talking points in turn:
• New GMO techniques are blurring the line between conventional breeding (implication: don’t even think about regulating or labelling the products of these techniques)
• GM foods are safe (an example of wishful thinking closing its eyes to the evidence: like Fox Mulder in the X-Files, the NAS ‘wants to believe’)
• GM foods and crops should be regulated by product, not process (implication: goodbye to Europe’s GMO regulations and labelling; hello to a US-style ‘light touch’ regulatory approach which assumes all GMOs are as safe as non-GMOs).
And the NAS report has worked like a GMO lobbyist’s dream, prompting parts of the British media, for instance, to lobby for Europe to unblock GM crop farming, on the grounds that the report’s eminent authors had blown away claims that GM crops and foods cause harm.
Perhaps not coincidentally, just two days after the publication of the NAS report, Health Canada announced its approval for human consumption of the world’s first commercialised GMO food animal, AquaBounty’s GM salmon. As GM foods have apparently been declared safe, who could possibly object?
The NAS report should be treated for what it is: not a scientific analysis, but a political document supporting US economic interests.
See our other coverage of the NAS report:
National Academy of Sciences GMO report fatally compromised by conflicts of interest
No, "science" has not confirmed that GMOs are safe to eat
And don’t miss Food & Water Watch’s report, Under the Influence: The National Research Council and GMOs.