GM foods are safe to eat
Studies show that GM foods can be toxic or allergenic
“Most studies with GM foods indicate that they may cause hepatic, pancreatic, renal, and reproductive effects and may alter haematological [blood], biochemical, and immunologic parameters, the significance of which remains to be solved with chronic toxicity studies.”
– Dona A, Arvanitoyannis IS. Health risks of genetically modified foods. Crit Rev Food Sci Nutr. 2009; 49: 164–1751
There are three possible sources of adverse health effects from GM foods:
- The GM gene product – for example, the Bt toxin in GM insecticidal crops – may be toxic or allergenic
- The GM transformation process may produce mutagenic effects, gene regulatory effects, or effects at other levels of biological structure and function that result in new toxins or allergens and/or disturbed nutritional value
- Changes in farming practices linked to the use of a GMO may result in toxic residues – for example, higher levels of crop contamination with the herbicide Roundup are an inevitable result of using GM Roundup Ready® crops (see Sections 4, 5).
Evidence presented below and in Sections 4 and 5 suggests that problems are arising from all three sources – throwing into question GM proponents’ claims that GM foods are as safe as their non-GM counterparts.
3.1.1. Feeding studies on laboratory and farm animals
Feeding studies on laboratory and farm animals show that GM foods can be toxic or allergenic:
- Rats fed GM tomatoes developed stomach lesions (sores or ulcers).2,3 This tomato, Calgene’s Flavr Savr, was the first commercialized GM food.
- Mice fed GM peas (not subsequently commercialized) engineered with an insecticidal protein (alpha-amylase inhibitor) from beans showed a strong, sustained immune reaction against the GM protein. Mice developed antibodies against the GM protein and an allergic-type inflammation response (delayed hypersensitivity reaction). Also, the mice fed on GM peas developed an immune reaction to chicken egg white protein. The mice did not show immune or allergic-type inflammation reactions to either non-GM beans naturally containing the insecticide protein, to egg white protein fed with the natural protein from the beans, or to egg white protein fed on its own. The findings showed that the GM insecticidal protein acted as a sensitizer, making the mice susceptible to developing immune reactions and allergies to normally non-allergenic foods. This is called immunological cross-priming. The fact that beans naturally containing the insecticidal protein did not cause the effects seen with the peas that expressed the transgenic insecticidal protein indicated that the immune responses of the mice to the GM peas were caused by changes in the peas brought about by the genetic engineering process. In other words, the insecticidal protein was changed by the GM process so that it behaved differently in the GM peas compared with its natural form in the non-GM beans – and the altered protein from the GM peas stimulated a potent immune response in the mice.4
- Mice fed GM soy showed disturbed liver, pancreas and testes function. The researchers found abnormally formed cell nuclei and nucleoli in liver cells, which indicates increased metabolism and potentially altered patterns of gene expression.5,6,7
- Mice fed GM soy over their lifetime (24 months) showed more acute signs of ageing in the liver than the control group fed non-GM soy.8
- Rabbits fed GM soy showed enzyme function disturbances in kidney and heart.9
- Female rats fed GM soy showed changes in uterus and ovaries compared with controls fed organic non-GM soy or a non-soy diet. Certain ill effects were found with organic soy as well as GM soy, showing the need for further investigation into the effects of soy-based diets (GM and non-GM) on reproductive health.10
- A review of 19 studies (including industry’s own studies submitted to regulators in support of applications to commercialise GM crops) on mammals fed with commercialised GM soy and maize that are already in our food and feed chain found consistent toxic effects on the liver and kidneys. Such effects may be markers of the onset of chronic disease, but long-term studies, in contrast to these reported short- and medium-term studies, would be required to assess this more thoroughly. Unfortunately, such long-term feeding trials on GMOs are not required by regulators anywhere in the world.11
- Rats fed insecticide-producing MON863 Bt maize grew more slowly and showed higher levels of certain fats (triglycerides) in their blood than rats fed the control diet. They also suffered problems with liver and kidney function. The authors stated that it could not be concluded that MON863 maize is safe and that long-term studies were needed to investigate the consequences of these effects.12
- Rats fed GM Bt maize over three generations suffered damage to liver and kidneys and alterations in blood biochemistry.13
- A re-analysis of Monsanto’s own rat feeding trial data, submitted to obtain approval in Europe for three commercialised GM Bt maize varieties, MON863, MON810, and NK603, concluded that the maize varieties had toxic effects on liver and kidneys. The authors of the re-analysis stated that while the findings may have been due to the pesticides specific to each variety, genetic engineering could not be excluded as the cause.14 The data suggest that approval of these GM maize varieties should be withdrawn because they are not substantially equivalent to non-GM maize and are toxic.
- Old and young mice fed GM Bt maize showed a marked disturbance in immune system cells and in biochemical activity.15
- Rats fed GM MON810 Bt maize showed clear signs of toxicity, affecting the immune system, liver and kidneys.14,15
- Female sheep fed Bt GM maize over three generations showed disturbances in the functioning of the digestive system, while their lambs showed cellular changes in the liver and pancreas.16
- GM Bt maize DNA was found to survive processing and was detected in the digestive tract of sheep. This raises the possibility that the antibiotic resistance gene in the maize could move into gut bacteria, an example of horizontal gene transfer.17 In this case, horizontal gene transfer could produce antibiotic-resistant disease-causing bacteria (“superbugs”) in the gut.
- Rats fed GM oilseed rape developed enlarged livers, often a sign of toxicity.18
- Rats fed GM potatoes showed excessive growth of the lining of the gut similar to a pre-cancerous condition and toxic reactions in multiple organ systems.19,20
- Mice fed a diet of GM Bt potatoes or non-GM potatoes spiked with natural Bt toxin protein isolated from bacteria showed abnormalities in the cells and structures of the small intestine, compared with a control group of mice fed non-GM potatoes. The abnormalities were more marked in the Bt toxin-fed group. This study shows not only that the GM Bt potatoes caused mild damage to the intestines but also that Bt toxin protein is not harmlessly broken down in digestion, as GM proponents claim, but survives in a functionally active form in the small intestine and can cause damage to that organ.21
- Rats fed GM rice for 90 days had a higher water intake as compared with the control group fed the non-GM isogenic (genetically the same except for the genetic modification) line of rice. The GM-fed rats showed differences in blood biochemistry, immune response, and gut bacteria. Organ weights of female rats fed GM rice were different from those fed non-GM rice. The authors claimed that none of the differences were “adverse”, but they did not define what they mean by “adverse”. Even if they had defined it, the only way to know if such changes are adverse is to extend the length of the study, which was not done. The authors conceded that the study “did not enable us to conclude on the safety of the GM food”.22
- Rats fed GM Bt rice developed significant differences as compared with rats fed the non-GM isogenic line of rice. These included differences in the populations of gut bacteria – the GM-fed group had 23% higher levels of coliform bacteria. There were differences in organ weights between the two groups, namely in the adrenals, testis and uterus. The authors concluded that the findings were most likely due to “unintended changes introduced in the GM rice and not from toxicity of Bt toxin” in its natural, non-GM form.23
- A study on rats fed GM Bt rice found a Bt-specific immune response in the non-GM-fed control group as well as the GM-fed groups. The researchers concluded that the immune response in the control animals was due to their inhaling particles of the powdered Bt toxin-containing feed consumed by the GM-fed group. The researchers recommended that for future tests involving Bt crops, GM-fed and control groups should be kept separate.24 This indicates that animals can be extremely sensitive to very small amounts of GM proteins, so even low levels of contamination of conventional crops with GMOs could be harmful to health.
In these studies, a GM food was fed to one group of animals and its non-GM counterpart was fed to a control group. The studies found that the GM foods were more toxic or allergenic than their non-GM counterparts.
3.1.2. Masking statistical significance through the concept of “biological relevance”
Study findings such as those described above have made it increasingly difficult for GM proponents to continue to claim that there are no differences between the effects of GM foods and their non-GM counterparts – clearly, there are.
To sidestep this problem, the GM industry and its allies have shifted their argument to claim that statistically significant effects, such as those found in the above studies, are not “biologically relevant”.
The concept of biological relevance was initially promoted by the industry-funded group, the International Life Sciences Institute (ILSI), and affiliates to argue against regulatory restrictions on toxic chemicals.25 But increasingly, it has been extended to the field of GM crops and foods.26 Biological relevance offers a route through which GM proponents can admit that feeding experimental animals a GM diet can cause statistically significant observable effects, but at the same time argue that these effects are not important.
However, this argument is scientifically indefensible. Biological relevance with respect to changes brought about by GM foods has never been properly defined, either scientifically or legally. Most feeding trials on GM foods, including those carried out by industry to support applications for GM crop commercialisation, are not long-term but medium-term studies of only 30–90 days long and therefore cannot thoroughly assess the safety of GMOs.
In order to determine whether changes seen in these medium-term studies are biologically relevant, the researchers would have to:
- Define in advance what “biological relevance” means with respect to effects found from feeding GM crops
- Extend the current study design from a medium-term to a long-term period. In the case of rodent studies, this would be two years – the approximate duration of their life-span11
- Examine the animals closely to see how the changes found in 90-day studies progress – for example, if they disappear or develop into disease or premature death
- Analyze the biological relevance of the changes in light of the researchers’ definition of the term
- Carry out additional reproductive and multigenerational studies to determine effects on fertility and future generations.
Since these steps are not followed in cases where statistically significant effects are dismissed as not “biologically relevant”, assurances of GM food safety founded on this line of argument are baseless.
In parallel with “biological relevance”, a trend has grown of claiming that statistically significant effects of GM feed on experimental animals are not “adverse”.27 However, the term “adverse” is not defined and the experiments are not extended to check whether changes are the first signs of disease. So again, the term is technically meaningless.
We conclude that GM proponents and regulatory bodies should cease masking findings of statistically significant effects from GM crops through poorly defined and scientifically indefensible concepts.
3.1.3. How misuse of “biological relevance” places public health at risk: Monsanto GM maize study
In 2007 a team led by Professor Gilles-Eric Séralini at the independent research institute CRIIGEN in France published a new analysis of a rat feeding study conducted by Monsanto with one of its GM maize varieties.
The maize, called MON863, was approved for feed and feed in Europe in 2005–2006.28 The maize was approved partly on the basis of the Monsanto study, which, however, could not be scrutinized by independent scientists and the public because the raw data were kept hidden on claimed grounds of commercial confidentiality. Only after court action in Germany forced disclosure of Monsanto’s data could Séralini and associates conduct their analysis.12
Séralini’s team found that according to Monsanto’s own data, rats fed GM maize over a 90-day period had signs of liver and kidney toxicity. Also, the GM-fed rats had statistically significant differences in weight from those fed non-GM maize control diets. The GM-fed females had higher concentrations of certain fats in their blood, and excretion of certain minerals was disturbed in GM-fed males.12
However, all statistically significant effects found in Monsanto’s study were dismissed by the European Food Safety Authority (EFSA) in its favourable safety assessment of the maize. They claimed that the statistically significant effects were not “biologically meaningful”.29,30 EFSA and GM proponents cited differences in response to the GM feed between male and female animals, claiming that toxic effects should be the same in both sex groups.11,31,32,33 However, this is scientifically indefensible as toxins with hormone-disrupting properties are well known to have different effects on males and females.34,35
Séralini commented on the dangerous trend of dismissing statistically significant effects by claiming lack of biological relevance in a 2011 review of the scientific literature assessing the safety of GM crops: “The data indicating no biological significance of statistical effects in comparison to controls have been published mostly by [GM crop development] companies from 2004 onwards, and at least 10 years after these GMOs were first commercialized round the world”. Séralini called the trend a matter of “grave concern”.11
After years of heavy criticism of the “biological relevance” tactic by independent scientists and a member of the European Parliament,36,11,37 in late 2011 EFSA issued an Opinion on the relationship between statistical significance and biological relevance.38
But EFSA’s Opinion failed to give a rigorous scientific or legal definition of what makes a statistically significant finding not “biologically relevant”. Instead, it allowed industry to come to its own conclusion on whether changes found in an experiment are “important”, “meaningful”, or “may have consequences for human health”. These are vague concepts for which no measurable or objectively verifiable endpoints are defined. Thus they are a matter of opinion, not science.
Moreover, the lack of a sound definition of biological relevance means that regulators have no strong scientific or legal grounds to disagree with industry’s claim that a statistically significant finding is not biologically relevant. This, in effect, makes GMOs impossible to regulate.
The conclusions of the EFSA Opinion are not surprising, given that it is authored by several affiliates of the industry-funded group, the International Life Sciences Institute (ILSI), including Harry Kuiper39 (also the chair of EFSA’s GMO panel), Josef Schlatter, and Susan Barlow.40 Because ILSI is funded by GM crop development companies, allowing ILSI affiliates to write EFSA’s scientific advice on how to assess the safety of GM foods and crops is akin to allowing a student to write his or her own examination paper – or allowing scientists to review their own papers submitted for publication!
3.1.4 Masking statistical significance through the concept of “normal variation”
Studies often find statistically significant differences in the composition of GM foods compared with their isogenic or near-isogenic non-GM counterparts (isogenic means genetically identical except for the one gene of interest, in this case the genetically modified gene). Studies also find statistically significant differences in animals fed a GM crop variety compared with animals fed the isogenic or near-isogenic variety.
However, GM proponents consistently dismiss these statistically significant differences in the experiment under examination by claiming that they are within the “normal variation range” or “within the range of biological variation”.
This tactic was used in a review of animal feeding studies on GMOs (the review included many of the studies summarised in this report). In spite of the significant differences found in the GM-fed animals, the reviewers used the concept of normal variation to argue that “GM plants are nutritionally equivalent to their non-GM counterparts and can be safely used in food and feed”.26
A confounding factor in some of these animal feeding studies was that the non-GM isogenic or near-isogenic variety was not made available to the researchers by GM developer companies. So the researchers had to do the best they could with the materials available to them. This meant using a non-GM crop variety as the comparator that was not the non-GM isogenic or near-isogenic variety.
Even given this limitation with some of the studies, it is scientifically unjustifiable to dismiss statistically significant changes in the GM-fed animals on the basis that they are within the normal range of variation. GM proponents define the “normal range of variation” by collecting values from many different studies carried out across a wide range of dates, using different experimental conditions and measurement methods. The result is a set of numbers that vary widely, but there is no scientific justification for including those numbers in the same dataset. On the contrary, there is much justification for excluding most of the values.
By using a dataset with such an unjustifiably wide range of variation, GM proponents are able to hide the genuine and meaningful differences between the GMO of interest and the valid controls – namely the isogenic or near-isogenic variety.
This is an attempt to minimize statistically significant differences brought about by the GM process by artificially widening the range of values compared beyond what can be scientifically justified. The practice runs counter to the aim of scientific experiments, which are designed to minimise variables. According to rigorous scientific practice, in any single experiment, the scientist manipulates just one variable in order to test its effect. In this way, any changes that are observed can be traced to a probable single cause.
In an animal feeding trial with GMOs, the manipulated variable is the GMO. One group of animals, the “treated” group, is fed a diet containing the GMO. Another group, the control group, is fed a similar diet, with the only difference being that it has not been subject to genetic modification. All conditions of the experiment outside the GM component of the treated group’s diet must be the same. Within this tightly controlled setup, any changes seen in the treated group are likely to be caused by the GM process.
Therefore, in any experiment to discover the effects of a GMO in an animal feeding trial, the only valid comparator is the control group within that same experiment (the concurrent control).
By comparing the treated group with a wide variety of control groups from other experiments (sometimes called “historical control data”), GM proponents are masking the effects of the GM process or GM diet, as any GM-related changes will disappear in the “noise” of the changes caused by many variables.
3.1.5. Regulators currently do not require long-term tests on GMOs
In order to detect health effects caused over time in humans eating GM foods, long-term (chronic) animal feeding trials are needed. But currently, no long-term tests on GM crops or foods are required by regulatory authorities anywhere in the world. Reproductive and multigenerational tests, which are necessary to discover effects of GM crops or foods on fertility and future generations, are also not required.11
This contrasts with the testing requirements for pesticides or drugs, which are far more stringent. Before a pesticide or drug can be approved for use, it must undergo one-year, two-year, and reproductive tests on mammals.12 Yet GM foods escape such testing, in spite of the fact that virtually all commercialised GM foods are engineered either to contain an insecticide or to tolerate being sprayed with large amounts of herbicide, so they are likely to contain significant amounts of pesticides.
The longest tests that are routinely conducted on GM foods for regulatory assessments are 90-day rodent feeding trials, and even these are not compulsory.11 While a 2012 EU draft regulation requests such tests for the time being, the wording is weak and foresees a situation in which they are not required.41 Also, the type of findings that would trigger a regulatory requirement for such tests has not been specified.42
Such 90-day rodent trials are medium-term (subchronic) tests that correspond to only a few years in terms of human lifespan and are too short to show long-term effects such as organ damage or cancer.43 In addition, too few animals are used in these industry tests to reliably detect harmful effects.
In spite of these serious shortcomings of regulatory tests, statistically significant harmful effects have been found even in industry’s own 90-day rodent feeding trials. The most common effects observed are signs of toxicity in the liver and kidney, which are the major detoxifying organs and the first to show evidence of chronic disease.11
These observations are consistently interpreted by GM proponents and regulators as “not biologically significant” or as “within the range of normal variation”, using the spurious arguments described in Section 3.1.4, above.
3.1.6. Stacked-trait crops are less rigorously tested than single-trait crops
Most GM crops currently on the market and in the approvals pipeline are not single-trait crops but stacked-trait crops. “Stacked-trait” means that several GM traits are combined in one seed. For example, GM SmartStax maize has eight GM traits: six for insect resistance (Bt) and two for tolerance to different herbicides.
Biotech companies have had to resort to developing multi-trait crops because of the failure of single traits. For example (see Section 5):
- Bt crops have fallen victim to secondary insect pests
- Pests have developed resistance to single Bt toxins
- Weeds have become increasingly resistant to glyphosate, the herbicide that most first-generation GM crops were engineered to tolerate.
Stacked GM crops present more of a regulatory challenge than single-trait crops because of the risk of unexpected interactions between the different GM genes introduced into the crop – and between the introduced GM genes and the genes of the host plant. There is also the risk of combination effects from toxins produced in the plant and/or pesticide residues. In short, the addition of multiple traits to a single crop increases the risk of unexpected and unintended harmful side-effects.
However, stacked-trait GM crops are even less rigorously investigated for possible health effects than single-trait GM crops. While the US does not require toxicological testing of any GM crops, Europe currently requires 90-day toxicological testing on single-trait GM crops. But in the case of stacked-trait crops, the EU food safety authority EFSA does not require toxicity testing of the final stacked-trait crop, believing that it can assess the toxicity of the final stacked-trait crop by looking at industry test findings on the single-event crops that were used to develop it.44
This move is irresponsible in the extreme, as such an assessment process depends on a series of assumptions, not on scientific testing. It fails to look at the actual effects of the mixed transgenes and their products within the crop.