Agency ignores and deceptively misrepresents scientific evidence. Report by Claire Robinson and Prof Michael Antoniou
In July the European Food Safety Authority (EFSA) published its opinion on the French health agency ANSES’s evidenced-based position of December 2023, which criticised Annex I of the European Commission’s proposal on the deregulation of plants obtained from new genetic modification techniques (new genomic techniques, NGTs).
According to Annex I, “category 1” new genetically modified (GM) plants (NGT1) can be treated as equivalent to conventionally bred plants – and not be subjected to the risk assessment, traceability or labelling requirements applied for the past decades to other GM plants – based on a few simplistic criteria, such as the number and size of the intended genetic modifications. Unintended modifications in the shape of mutations (DNA damage) created by the GM processes used are ignored in the Commission’s criteria.
In a move that came as no surprise, EFSA agreed with the Commission, concluding that new genetically modified organisms (GMOs) can be considered equivalent to conventional plants if they correspond to the Annex I criteria. However, this opinion disregards important scientific evidence and seems to be designed to deliver a politically and economically convenient result to suit the interests of the agricultural biotechnology lobby, rather than to protect public health and the environment.
In particular, EFSA uses the wrong comparison to justify exempting NGT-derived plants from regulatory safety checks, comparing NGTs to chaotic chemical- or radiation-induced random mutagenesis breeding instead of conventional breeding. This is like comparing the risks of a highly toxic new pesticide to those posed by an older toxic pesticide and concluding that since the older pesticide is already out there, we may as well release the newer pesticide without regulatory oversight.
EFSA also misuses scientific evidence on the DNA damage caused by random mutagenesis, seemingly in order to “normalise” the DNA damage caused by NGTs and make it more publicly acceptable. This is in stark contrast to global regulatory policies governing (for example) chemical and radiation exposure, which aim to protect all life from the highly damaging effects resulting from exposure to DNA mutagens.
The following analysis explains the serious faults in EFSA’s arguments.
What does ANSES say?
ANSES correctly points out that it’s not just the size or number of the genetic modifications that is important, but what they do – their functional consequences. Knowing the size and number of intended mutations tells you nothing about this.
These consequences are explained in a further position statement published by ANSES on the health, environmental, and socio-economic risks of NGT-derived GMOs. Based on 10 case studies of existing NGT-derived GM plants, ANSES wrote that “certain potential risks appear repeatedly in these case studies” and that “These include risks linked to unexpected changes in the composition of the plant, which could give rise to nutritional, allergenicity or toxicity problems, or medium- and long-term environmental risks, such as the risk of gene flow from edited plants to compatible wild or cultivated populations.”
GMWatch has repeatedly drawn attention to the potential and actual “functional consequences” of genetic manipulations in NGT-derived GMOs, based on a large body of scientific studies and reviews.
EFSA momentarily agrees with ANSES's point, but then says that we also don’t understand the “functional consequences” of genetic variations in conventional plants, so they see no extra risk from GM plants obtained via NGT1 methods.
In its opinion, EFSA ignores scientific publications showing that NGTs can create mutations that are different from those arising naturally, from conventional breeding, or even from random mutagenesis breeding (where plant material is subjected to radiation or treatment with chemicals to induce mutations). Therefore, scientists have warned that the mutations arising from the application of NGTs can present very different and more serious risks compared with conventional or random mutagenesis breeding.
We would welcome EFSA’s specific arguments against the findings of these publications. We would feel reassured by the fact that they had engaged with them and found evidence-based reasons to reject them. But the agency simply pretends they don’t exist and doesn’t include them in its limited list of references, in a case of “see no evil, hear no evil, speak no evil”.
Natural mutations aren’t necessarily safe
In addition, even if it could be proven that the mutations in NGT1 plants are of the type that could happen naturally, this doesn’t mean it’s acceptable, safe, or desirable to deliberately create them rapidly and on a large scale using NGT methods, without robust regulatory oversight. Certain types of genetic mutation can occur naturally, but they can cause health- and life-threatening diseases and deformities in people, animals, and plants. So we must treat technologies aimed at creating such mutations with great caution.
The argument, “If nature does it, it’s OK for us to do it too” is full of holes, though it’s far from the first time that it’s been used. Previously, agricultural GMO lobbyists have stated that because an aspect of genetic engineering called horizontal gene transfer (HGT – transferring genes across species) has occurred naturally across species in a variety of sweet potato at some point in the plant’s evolutionary history, it’s acceptable and safe for humans to expedite HGT in genetically engineering plants to contain foreign genes – and for the products to avoid any special safety assessment.
Prof Jack Heinemann said in reply to such claims: “Of course nature can also create organisms – by HGT or other means – that are capable of causing us harm. But that is no reason for us to do it unwittingly to ourselves. Nature can squash us with a rock from space, causing injuries indistinguishable from a car crash. This is not a reason to stop motor vehicle safety testing or recommend removing seat belts.”
Prof Heinemann and co-authors formalised their arguments in a peer-reviewed publication, which explained:
“Language such as ‘natural GMOs' or ‘naturally transgenic’ (Kyndt et al., 2015) instead of ‘spontaneous/natural horizontal gene transfer’ still demonstrates – and causes – confusion, as in linking GMOs to the observation of sweet potato cultivars containing DNA sequences also found in a bacterial pathogen of plants. As the journal Nature proclaimed: ‘The sweetpotato genome contains genes from bacteria, so is an example of a naturally occurring genetically modified (GM) plant’ (Anon, 2015). In fact, all plants and animals have genes from viruses and bacteria. Some come from mitochondria that reside in each of our cells. Mitochondrial ancestors were free-living bacteria. Some of these genes from the mitochondria have migrated to the chromosomes kept in the cell nucleus (Martin, 2003). There are many examples of genes from viruses and bacteria in plant and animal genomes because of horizontal gene transfer. It would have been extraordinary if the sweet potatoes had no DNA from bacteria.
“These headlines and phrases linking spontaneous events in nature to outcomes of gene technology have the purpose to ‘influence the public’s current perception that transgenic crops are ‘unnatural’” (Kyndt et al., 2015). The DNA that the researchers found spontaneously had entered into the sweet potato lines via the same bacterium that was, beginning in the 1980s, manipulated by technologists to deliver transgenes into plant cells. Presumably, they thought that the distinction would be lost on nonbiotechnologists, in the same way that someone might confuse getting poked by a sharp branch while walking through a forest with being pierced by an arrow shot from a bow, because both the branch and arrow are made of wood.”
EFSA denies the obvious
In its opinion, EFSA engages in a spectacular exercise of denial that the Commission’s proposal actually means what it clearly says.
EFSA correctly says the proposal “defines criteria to classify NGT plants as ‘category 1 NGT plants’, equivalent to conventional plants, or NGT2, not equivalent to conventional plants.” However, EFSA then adds, “These criteria are not meant to define levels of risk but allow certain NGT plants to be classified as equivalent to conventionally bred plants.” It repeats this message several times in its opinion.
EFSA is wrong in saying that the criteria don’t say anything about risk. The Commission makes clear that the category 1 criteria do define the assumed level of risk. For example, the Commission says, “‘Category 1 NGT plants’... should be treated as plants that have occurred naturally or have been produced by conventional breeding techniques, given that they are equivalent and that their risks are comparable” (our emphasis). The Commission deduces that as category 1 NGT plants (supposedly) pose no extra risks, they are exempted from the requirement for the pre-market risk assessment that currently applies to all GM plants.
Why would EFSA claim that the Commission’s criteria don’t define the level of risk, when the Commission clearly says they do? Is it because EFSA is trying to distract attention from the undoubted outcome of these criteria – that human and animal health and the environment will not be protected from the real risks that NGT-derived GMOs present?
Perhaps because EFSA can foresee how this Alice-in-Wonderland reasoning will be received, it then instantly reverts to the question of risk, but only to deny that NGT plants (apparently of both categories, 1 and 2) pose any risk at all: “Moreover, with respect to the potential risks from NGT plants, the EFSA GMO Panel did not identify any additional hazard associated with the use of NGTs compared to conventional breeding techniques, which include random mutagenesis using physical or chemical agents.”
This is a strange and scientifically indefensible statement to make about NGT-derived plants as a whole, given the extreme genetic changes that can be brought about using new GM techniques and the widescale unintended changes that accompany the intended ones.
Scientists estimate that around 94% of new GM plants affected by the Commission’s proposal would be classed as category 1 according to its inadequate criteria, even though “our analysis suggests that they could bear environmental risks comparable to those of other GMOs, including potential insecticidal NGT1-plants based on RNA interference (RNAi) technology”.
When scientific bodies make statements that defy all logic and scientific principles, it is an indication that they are either technically out of their depth, incapable of rational thought, or attempting to deceive the public. Any of these explanations gives cause for concern.
EFSA misrepresents what ANSES said
EFSA also misrepresents what ANSES said. To understand this, we need to revert to the Commission’s original text.
The Commission proposes that “Deletions and inversions of any number of nucleotides” – the smallest unit of DNA and RNA – brought about by gene editing are acceptable for category 1 NGTs. In other words, the genetic engineer can create deletions or inversions of any number of nucleotides in a new GMO and it still won’t be subjected to safety checks, traceability, or labelling requirements.
ANSES commented that in conventionally bred plants, “pan-genome analyses... show that the size distribution of structural variants is strongly biased in favour of sizes of around one kilobase or less. Large variants exist but are much less common... structural variant-type deletions observed in conventional plants are usually limited to around one kilobase in size.”
EFSA applies this comment to genomic deletions, saying, “ANSES claims that the available scientific literature shows that [in conventionally bred varieties] genomic structural variations have a size distribution of the order of a kilobase [1000 DNA base units] or less. However, the occurrence of larger deletions during plant evolution is well documented, even for cultivated crops generated by conventional breeding.” EFSA references several scientific papers to back up its argument that larger deletions happen in conventionally bred plants and to support its view that their occurrence in new GM plants is not of special concern.
Case closed? Not quite. Contrary to EFSA’s claim, ANSES doesn’t say all structural variations in conventionally bred plants are smaller than 1000 bases. It says there’s a trend whereby most deletions in conventional plants are not as large as the deletions that the Commission is proposing can be present in a category 1 NGT plant and it still be considered conventional-like.
The occurrence of larger deletions in conventionally bred plants might, as EFSA says, be “well documented”, but it might still be relatively rare, which is what ANSES is saying.
And crucially, EFSA doesn’t go on to state what the “functional consequences” of such large deletions in conventional plants are. Is the resulting plant deformed or infertile, or does it have potentially dangerous compositional changes? We don't know, and EFSA apparently doesn’t want to know.
The effect of EFSA’s misrepresentation of ANSES’s statement is to paint a false picture of conventional breeding as more damaging to the genome – and gene editing as less damaging – than they really are.
Random mutagenesis creates freaks
EFSA references several scientific papers (Bolon et al 2014; Li et al 2001, 2016; Liu et al 2023; Morita et al 2009) in order to present the idea that large deletions and inversions occur in “conventional breeding”, thus normalising these mutations in our minds. But in many cases these papers actually record genetic freaks – defective and abnormal plants made by random mutagenesis breeding.
Random mutagenesis breeding is not conventional breeding. In fact, EU law defines it as a GM technique, albeit one that isn’t subject to the requirements of the GMO regulations due to its long history of use. It is a genetically disruptive technique used alongside a minority of conventional breeding. The products of random mutagenesis breeding were correctly described by the scientist Dr Angelika Hilbeck as “freaks” in a recent webinar.
Mutagenesis breeding is a decades-old technique in which seeds or other plant materials are exposed to radiation or chemicals in order to create large numbers of mutations (DNA damage). Plant breeders hope that in rare instances, one or more of these mutations might turn out to confer a useful trait. If they find a mutant plant with such a useful trait, they breed on from it – if it also turns out to be fertile, which may not be the case.
Random mutagenesis breeding is known to be highly risky to the plant, creating large numbers of deformed, non-viable and infertile plants. Thousands or millions of undesirable plants are discarded in order to identify plants suitable for further breeding. Is random mutagenesis breeding risky for the human or animal consumer? No one knows, as the necessary safety research hasn’t been done.
It is not surprising, then that random mutagenesis breeding had its heyday in the 1980s and 1990s and has largely fallen out of use since then. The IAEA database of mutant varieties (which includes ornamental plants and conventionally bred offspring of mutant varieties) only contains around 3000 records – a tiny fraction of the quarter of a million crop varieties available in the world.
New GM creates new risks
Even though random mutagenesis excels in creating “freaks”, EFSA is wrong in assuming that the risks of NGTs are no greater. Scientific reviews show that new GM techniques can create both intended and unintended changes in the genome that would not occur, or would be extremely unlikely to occur, from random mutagenesis methods.
The reason is that in conventional breeding and random mutagenesis, some regions in the genome undergo changes less frequently than others because these regions are protected by DNA repair mechanisms in the cell. In contrast, the gene-editing tool CRISPR/Cas can bypass these naturally occurring protections, meaning it can access parts of the genome that are not accessible to change in conventional and random mutagenesis breeding.
The key to the different risk profiles of conventional breeding versus new GM techniques lies in the fact that the mutations that occur in conventional breeding of plants are NOT random and directionless, as shown in important research. This research shows that genetic variation arising from rounds of natural reproduction is biased in a “directed evolutionary” manner that benefits the plant.
It’s possible to draw the conclusion from the study that genetic evolution doesn’t just happen through chance but is driven by an as yet undefined more intelligent mechanism.
While this study was done on conventionally bred plants, the results would also apply to random mutagenesis bred plants, which, in spite of the name, causes mutations that are not entirely random but, as Testbiotech has pointed out, are still under the control of the plant’s own gene regulatory processes.
In contrast, in the case of new GM techniques – which are not involved or mentioned in this study – the unintended DNA damage that results from the processes used is random, across the whole genome, with no regions protected. This is both the “unique selling point” of new GM techniques – they can access parts of the genome that conventional breeding and mutagenesis breeding cannot – and a central reason for its higher risk profile.
EFSA ignores these important reviews and their conclusions, turning a blind eye to years’ worth of scientific evidence with the apparent aim of delivering the politically acceptable yet false verdict that NGTs do not create new risks beyond those of “conventional breeding” – its deceptive euphemism for random mutagenesis.
Long history of safe use?
While EU law refers to random mutagenesis’s long history of use, can we say it’s a long history of safe use?
That's a debatable point. We know random mutagenesis breeding is risky for the plant that is mutagenised, but we don’t know if it’s risky for the consumer or the environment as the tests haven’t been done. We are somewhat protected by the fact that it’s inefficient and has largely fallen out of use, so there simply aren’t that many mutagenised food crops out there. Those that are out there have often been crossbred so much with conventional plants that any harmful-to-us mutations will have been diluted out.
It is disingenuous of EFSA to compare new GM techniques such as gene editing with random mutagenesis. It is the wrong comparison. EFSA is placing new risky and imprecise technologies (NGTs) next to an older risky and imprecise technology (random mutagenesis) and is saying, in effect, that new GM creates similar messes to random mutagenesis and therefore we should give it a free regulatory pass as it's no worse. By this logic, regulation becomes a “race to the bottom” – a levelling down to match or exceed the worst standards we've allowed in the past.
If random mutagenesis breeding were invented in the early 2000s instead of in the 1920s, when we knew very little about the structure and function of the genome, we would not have introduced plants developed with it without far more careful scrutiny. Yet EFSA, a hundred years later, is using this chaos-creating technique as a flimsy excuse to give another even more chaos-creating technique a free pass into our fields and food supply.
Large deletions and inversions can’t be assumed to be OK
What of the scientific papers that EFSA references to suggest that large deletions and inversions of DNA segments are normal in conventional breeding? Here again, EFSA misleads us. One of the papers (Seah S et al 2004) shows that when a large inversion was found in a conventionally bred tomato, the tomato was susceptible to attack by nematodes (organisms that feed off and damage the roots). Tomatoes without the inversion were resistant to the nematodes. Such a genetic fault could make the difference between commercial failure and success of a crop.
In other words, EFSA is pointing to a damaging genetic abnormality that could adversely impact the plant's health and survival as being perfectly acceptable in a new GMO plant derived from NGTs. It believes that any defective products of a technology prone to creating such mutations do not need to be subjected to risk assessment, traceability, or labelling – and that they should be rolled out on a large scale in farmers’ fields and eaten by human and animal consumers without regulatory oversight.
How EFSA misuses evidence: The detail
The other papers that EFSA cites to reassure us about the use of NGTs and normalise their mutational activity actually show how dangerous and risky these techniques are, in terms of the massive damage they cause to the plant’s genome. Below a critique is provided of the papers EFSA uses in support of its dismissal of ANSES’s position. These points are fairly technical but are offered for those who need this degree of detail.
* Bolon YT et al 2014: This paper describes the extreme types of mutation caused in soybeans by radiation-induced random mutagenesis, including deletions, “surprisingly high rates” of duplication, a “tandem duplication” (duplications next to one another), and a translocation. The authors did genetic mapping to investigate the functional consequences of some of the mutations. They found a change in seed composition (higher seed oil and lower seed protein levels) for one mutant and short leaf stalks in another mutant.
GMWatch comments that these mutations in themselves are not necessarily dangerous for consumers or risky for farmers, but they might be. Only specialised tests could tell for certain.
* Li X et al 2001: This paper on radiation-induced random mutagenesis describes how researchers created large numbers of mutations by bombarding rice and Arabidopsis (a model plant often used in GM experiments) seeds with radiation. Then they developed a method to identify the different deletion mutations in the plants.
GMWatch comments that the research was not designed to compare mutation frequency or type in mutagenesis breeding, conventional breeding, or new GM techniques, let alone the safety of these mutations or their functional effects on the organism. EFSA only seems to be citing it to claim (falsely) that “conventional breeding” creates similar mutations to new GM.
* Li G et al 2016: This is yet another paper showing the massive DNA damage caused by radiation-induced random mutagenesis, in this case in rice. The radiation-induced mutations observed ranged from DNA strand breaks and damage to base groups to single base substitutions (SBSs), deletions, insertions, and duplications.
Interestingly, the authors calculated that on average only 9.6 spontaneous mutations (7.6 SBSs and two small deletions/insertions) occurred spontaneously in rice in third-generation conventionally bred offspring of mutagenesis bred rice. The authors noted that these spontaneous mutations, which were typically small in size, only “slightly inflate” the number of mutations caused by the radiation. This means that the spontaneous mutations, compared with the radiation-induced mutations, are very small in number.
In comparison, the number of mutations in a radiation-induced mutagenised rice line was much greater – between 28 and 78, with an average of 59. As a reference comparison from other studies, there were 1.4 insertions per rice line in the transgenic (T-DNA) GM insertion mutant collection created in the rice variety Dongjin and around 390 SBSs per line in a Nipponbare chemically-induced mutagenised rice collection. In Li G et al 2016, the number of genes affected in each of the 41 third-generation conventionally bred offspring lines of mutagenesis bred rice lines ranged from seven to 147, with an average of 31. The large variation in mutated genes per line was due to the presence of large deletions.
The authors conclude that the radiation caused “more deleterious mutations” in the first mutagenised generation of rice than they had expected. Unsurprisingly, they also found that 27% of the 10,000 first generation lines were sterile.
GMWatch comments that the study shows that radiation-induced mutagenesis is highly damaging to the genome and that the number of mutations arising from this method is much greater than those arising from conventional breeding. The findings confirm those of other rice research that shows vast numbers of mutations arising from tissue culture and plant cell transformation process – obligatory steps of all genetic engineering procedures, both new and older-style – versus a far smaller number from conventional breeding. It also shows that the types of mutations are different, with radiation-induced mutagenesis tending to create larger mutations than those arising from conventional breeding. It further shows a survival-threatening consequence of such mutations – sterility.
It is a deceptive misuse of evidence to deduce from this study, as EFSA does, that large deletions and inversions of DNA segments are normal in “conventional breeding” and that new GM technologies highly prone to creating such mutations are no more risky than conventional breeding.
* Liu J et al 2023: This paper concerns a genetic variation called tandemly arrayed genes (TAGs, in which one gene is duplicated and the copy is found adjacent to the original). It shows that CRISPR/Cas gene editing in plants can create targeted TAG deletions and that these deletions can result in deletion-inversion mutations called delinver mutations. The authors state, “Our work raises the alarm about delinver mutations as common unwanted products of targeted TAG deletions in plants”. They raise the question of whether this will also happen in humans subjected to gene therapy using the CRISPR gene editing tool.
These large chromosomal rearrangements will affect many gene functions as well as alter gene copy number (gene dosage). And gene dosage matters a lot with respect to health and disease. Too high a dosage, which can be caused by gene duplication, can result in problems such as heart disease, cancers and neuropsychiatric disorders.
In plants, such genetic rearrangements could change biochemical pathways and result in unexpected toxicity or allergenicity.
While Liu et al correctly state that TAGs can exist naturally in humans and plants (and no doubt this is why EFSA uses this paper in support of its arguments), the authors do not make the error of concluding that this means they are safe or desirable. It is obvious that long processes of evolution have selected to maintain mutations or genetic variations that don't cause harmful effects and/or help fitness. Those that cause harmful effects would have been selected out over time.
But with genetic engineering, we don't have that protective element of time. Mutations – including extreme ones – will be rolled out on a wide geographical scale and in a short time scale, as Prof Jack Heinemann has pointed out. This is why all such mutations must be evaluated for their effects in dedicated risk assessments.
It is inexcusable for EFSA to use the findings of this worrying paper to normalise the creation of mutations via NGTs and to claim that such mutants do not require any more scrutiny than the products of conventional breeding.
* Morita R et al 2009: This study shows that radiation-induced mutagenesis causes small and large deletions and inversions in rice. The mutations observed cause some undesirable effects, such as overproduction of leaves, protein deficiency in the grains, chlorophyll deficiency (which causes impaired photosynthesis), and dwarfism (which can be seen as desirable or undesirable, depending on the aims of the breeder).
GMWatch comments that the findings show the damage that radiation can cause in the genome. That doesn’t mean it’s OK, safe, or desirable to deliberately create damage to the genome via new GM techniques and then enable the products to evade risk assessment.
Conclusion
EFSA misuses and ignores a large amount of scientific evidence to reject ANSES’s precautionary report and to justify its own stance that highly genetically disruptive new GM techniques do not need to be regulated. It makes a misleading comparison between NGTs and another genetically disruptive technique, random mutagenesis breeding, to falsely minimise the risks of new GM and maximise the risks of what it terms “conventional breeding”, though random mutagenesis breeding cannot be equated with conventional breeding, either in its processes or its effects.
Having said that, however, EFSA fails to recognise the proven differences between conventional and random mutagenesis breeding versus new GM techniques. These differences mean that new GM techniques pose new and potentially greater risks than conventional and random mutagenesis breeding.
EFSA also – in stark contrast with ANSES – fails to consider the functional consequences of the mutations caused by these different techniques, whether those consequences be for the plant itself, for consumers of the plant, or for farmers and wildlife.
EFSA’s opinion appears to aim at delivering a politically and economically motivated “rubber stamp” of approval for new GM techniques that suits the interests of the agricultural GMO lobby, rather than a scientifically based and precautionary opinion that would protect the public and the environment from the risks and harms that could arise from the application of these NGTs.
EFSA should revise its opinion by honestly addressing the evidence presented in this article and all other relevant evidence in the peer reviewed scientific literature.