The hidden side of clinical trials

Watch the AllTrials TEDx talk on YouTube

Learn more

Evidence matters to the public

Join us on 1st November at Parliament to make the case

Learn more

Plant Science Panel

Insecticides, biofuels, GMOs …

Learn more

'The Ugly Truth'

by Tracey Brown, director of Sense About Science

Learn more

How useful are animal feeding studies in plant research?

How useful are animal feeding studies in plant research?

Over the past year quite a few people in plant research have said they are not convinced that animal feeding studies show anything useful about a new crop; which would make them a waste of time, money and animals. But in the recent discussions following the retraction of a paper about feeding GM maize to rats, there was some heated discussion on twitter, blogs and forums about the need for more of them, for longer studies and for GM crops in particular.

In the EU, 90 day feeding trials on rodents are now mandatory for all GM crops. In some cases, a one year feeding trial is now required. Both of these decisions are scientifically controversial. Some people are in favour of more stringent requirements and longer feeding studies, others think we need fewer of them – for non-GM crops, there is no requirement to look at biosafety considerations at all. What actually happens and what are they for? Do you know everything you want to know about them?

On Wednesday 15th January 2014 our panel of experts Professor Joachim Schiemann, director of the Institute for Biosafety in Plant Biotechnology and coordinator of the EU-funded GRACE project (GMO Risk Assessment and Communication of Evidence), Professor Pablo Steinberg, director of the Institute for Food Toxicology & Chemical Analysis and Dr Wendy Harwood, senior scientist at the John Innes Centre, answered your questions on animal feeding studies in plant research.

Read our Storify about the event here.

If you have a question on a plant science related issue then get in contact with our plant science panel via Twitter, @senseaboutsci using #plantsci, or email us at [email protected]

26. Am I right that a 90-day study is done first and if toxicity signs are seen, only then does one plan a long-term (1- or 2-year) feeding study which could inform about chronic toxicity? Arguably, toxic effects may not become evident after such a short study period, is a mandatory 90-day study sufficient?  And how many times should a study be repeated before its deemed “safe” (low relative risk) for humans? (@MattOldach)

PS: "Yes, you would first do the 90-day study and then, if any toxic effect is seen, one would go for a longer study. Many toxicologists, including myself, doubt that a 90-day  trial is sufficient to see any toxic effect. However, one has to also take into account an ethical aspect: the very high amount of animals needed for a carcinogenicity study. This is why nobody would suggest performing the carcinogenicity study from the very beginning, but instead to first perform a 90-day study and then, only if symptoms of toxicity are seen, perform the long-term feeding trials."

"Regarding the question, “how many times should a study be repeated before its deemed “safe” (low relative risk) for humans?”, the clear answer is that if the study design is correct (and among other things, the number of animals to be used per group is adequate) the study is performed once and does not need to be repeated."

25. The current controversy and debate indicates that regulators should review their requirements for animal feeding studies as part of GMO risk assessment, taking into account the necessity and justification for these trials. What is the mechanism for the regulators to undertake such reviews? (Research Animals Department, RSPCA)

JS: "I fully agree with this statement. In the case of the 90-day feeding studies which became obligatory recently for GM food/feed approval in the EU, the European Commission will review its position at the end of 2015 in light of the outcome of GRACE and other credible scientific knowledge. By linking GRACE (which includes feeding trials, in vitro studies, omics and in silico analyses) with the publicly funded feeding trials in France and a forthcoming EU-funded project to perform 2-year feeding trials, we aim to provide a comprehensive set of unbiased data in a highly transparent way as a basis for future regulatory decisions."

24. Which food types/additives legally require toxicology studies in the UK? If there is an effective system for these, can we just apply it for genetically modified foods? (Rebecca Nesbit - @RebeccaNesbit)

PS: "Food additives must be assessed for safety before they can be used in food. European Union legislation requires most additives used in foods to be labeled clearly in the list of ingredients, with their function, followed by either their name or E number. An E number means that it has passed safety tests and has been approved for use in the UK and in the rest of the EU."

"The current procedures for evaluation and authorisation of genetically modified foods are laid down in Regulation (EC) 1829/2003 on GM food and feed, which came into force in April 2004. The safety assessments are carried out by the European Food Safety Authority (EFSA), according to its published guidelines for the assessment of GM plants. For more information see e.g. the Food Standards Agency page.  We cannot directly apply the food additive safety assessment to GM crops. Nowadays we have recommendations from OECD and EFSA on how to test GM crops and this should be sufficient: research projects funded by the European Union such as GRACE will show this in the near future."

23. Are animal feeding trials necessary to regulate GMO crop safety for regulatory approvals? Or freedom to operate? Very different purposes. (@ErikJanus)

WH: "In the EU, 90 day feeding studies are now required but there is a strong argument that they are not needed in many cases in order to regulate GM crop safety. Freedom to operate is a very different issue and should be considered separately from safety issues."

22. As companies are required to do testing before providing a product to the public, who should do them? Independent academics? Industry? Isn't there a conflict of interest?

PS: "I do not know of any regulatory agency with independent academics which would take over such a task. These studies are very expensive and I see no alternative to industry taking over the performance and costs of the testing. They are the ones wanting to bring the product to market and they should be very critical of testing results. If one product showing toxicity in a trial was brought to market and health problems arose in humans, then the product and the company would really have a problem. I do not think that any serious company would want to cope with such a  problem."

21. Could controversy over GM safety have been avoided by having sound regulation governing who is and who is not competent to carry out scientifically valid and ethically sound feeding trials on laboratory animals? (Dr David J James)

WH: "The peer review system would, in most cases, prevent the publication of flawed studies, however, there are sometime failures as in the Seralini paper. In addition to this, in most countries there would be strict guidelines governing work involving laboratory animals so additional regulation is probably not needed."

PS: "It is impossible to determine a priori who is competent and who is not. Any scientist can theoretically plan and carry out such a study. My experience says that many of those planning and carrying out those studies have not got or have insufficient experience and know-how to perform the studies properly. Moreover, if one adheres to international test guidelines, studies are usually extremely expensive, so that only a few scientists at all would be able to perform such studies."

20. Are these studies really the right way to answer the relevant safety concerns? (Johnna Roose)

PS: "At the moment we do not have better alternatives. The feeding trials will give us adequate answers, but one must bear in mind that the outcome of a feeding trial with one GM crop will not tell us what might happen with another. This means that we will have to look for alternatives in the long range, for example look for biomarkers of toxicity in animals being fed for 90 days with a GM crop that will eventually tell us something about the long-term toxicity of the crop, a project just being started by a French consortium."

WH: "There are now many different technologies that can be used to analyse GM crops to address safety concerns. These technologies continue to rapidly develop and improve and as this happens the justification for animal feeding studies is less strong."

19. If GM food is safe, why do we need to label it? Is it possible that some GM foods are coming onto the market unlabeled? (Claire Hastings)

WH: "GM food is labelled to give choice to consumers. Without labelling this choice would be removed. Labelling rules vary in different countries, for example in the EU labelling of GM food is mandatory but it is not in the US, although labelling is currently being discussed in different states."

18. One letter to the editor following Seralini’s paper said that similar studies published in the journal had used the same strain of rats. Is this the case, and if so should we be highlighting more studies which are unreliable? (Rebecca Nesbit - @RebeccaNesbit)

WH: "The strain of rats used in Seralini’s study might have been suitable for another  study that was perhaps shorter-term. The problems with the Seralini study do not mean that other studies using the same rats will also be unreliable."

PS: "There are a number of studies that do not comply with internationally accepted test guidelines and that describe results which cannot be used for risk assessment. With the documents published by the OECD and by EFSA, more valid studies hopefully will be published, so that one can decide on a serious basis if a new risk assessment is really needed for a certain GM crop."

17. How can serious mistakes like using the wrong strain of rat (such as Seralini) be avoided in future feeding trials? (@animalresearch)

PS: "By using a strain of rat such as Wistar, which shows a low incidence of non-neoplastic and neoplastic lesions. A rat strain that shows none of these signs does not exist. It is the same with human beings: with increasing age, tumours appear in a certain number of animals. One has to choose a strain in which this number is low."

JS: "Such mistakes can and have to be avoided by selecting the proper rat strain when planning the experiment. An important help to select the proper strain is to look into the historical data of the respective strain. For example, historical data for Wistar Han RCC are publicly available under"

16. "Regarding the Seralini GMO study, is there any information available on how much of the glyphosate stays with the GMO plant post-harvest? Further, with many GM-food crops having protective layers over the seeds (e.g. leaves over the corn cobs), how much transfer of residual glyphosate could be occurring? Are the levels of glyphosate being tested consistent with levels that could be entering the food system?" (Tommy Reagan)

WH: "The minimum level of glyphosate used in the Seralini study was chosen to represent a level that could be found in some contaminated tap water, the maximum was half of the working agricultural dilution.  The range tested was very large and the higher level chosen is likely to be very much higher than could be entering the food system. I am not aware of data showing whether the lower level was consistent with levels that could possibly be entering the food system."

15. Should feeding trials be multi-generational? (David Norton)

PS: "Feeding trials are not usually multi-generational. A multi-generation study would only be included if there is any sign that reproductive organs are affected by the material being tested."

14. Does it matter which strain of rat you use?

PS: "Yes it matters. One should use a rat strain with a low incidence of spontaneous tumors."

13. Why is it necessary to define a study before starting? What do you do if your toxicity study turns out to be a carcinogenicity study and you haven’t got enough animals in your groups? (Lydia Le Page)

PS: "Defining a study before starting is necessary to adequately design it (this means that you have to define the number of animals per dose to be used, the number of doses to be tested, etc. and the number of animals used per group varies with the length of the study). Usually one would plan a 90-day study if any sign of toxicity was seen in an acute or subacute toxicity study. If toxicity was seen in a 90-day study, then one would plan a long-term (1- or 2-year) feeding study. Since the end of last year the European Commission decided that in the case of GM crops, a 90-day feeding study is mandatory." 

"On the 2nd question, if you don’t have enough animals, then your study design is not correct and the results to be obtained cannot be used for a risk assessment of the genetically modified crop you’re testing. If you use fewer animals per group than indicated by internationally accepted guidelines, then you will obtain results that cannot be adequately analyzed with statistical methods, so you will not be able to say if there are any differences (for example in the number of tumors) arising in control and GM crop-fed animals."

12. Are GM crops cooked before being fed to animals in safety studies? Would this affect outcome? (Krishna Thaker)

PS: "GM crops are definitely not cooked before being fed to animals in safety studies. GM crops are digested as normal feed in the gastrointestinal tract of the animals, so that cooking before feeding would not affect the outcome of the safety studies."

11. Is there evidence that GM crops are dangerous? (Richy Stokoe - @Richy_Stokoe)

WH: "When considering risks associated with GM crops there are two main areas, food / feed safety and environmental safety. In terms of food / feed safety there is no good evidence that there are additional risks from eating GM over eating non-GM. In terms of environmental safety, this must be approached  on a case by case basis. It would not be sensible to grow certain crops in certain locations because of the risk of environmental impact but in other locations it would be fine."

10. Are animals really a good model for feeding studies? Would it be better to pay humans to volunteer to be involved? (Jan Hocking)

JS: "Rodents (rats and mice) are good models for testing the toxicity of chemicals. Their usefulness to evaluate toxic effects of whole GM food/feed is under debate. To perform these toxicology tests with humans would be unethical."

WH: "A difficult question where the answer depends partly on the specific animal chosen. I think the key word is ‘model’. These studies can only provide guidance as to what might happen in humans. With many animal models things like life span are very different to that in humans. As for human volunteers, well humans do volunteer for medical trials but with feeding studies there would be lots of problems to overcome."

9. @senseaboutsci #plantsci How is the predictive validity of animal feeding studies evaluated and assessed? (@RSPCA_LabAnimal)

PS: "At the present time the predictive validity of animal feeding studies for GM food/feed cannot be evaluated. This is because the animal feeding studies available in the public domain do not comply with internationally accepted guidelines."

8. How many animals are needed in a feeding trial for the results to be significant? Does it make a difference what chemical you're testing? (David Norton)

PS: "It depends if you want to study mid- or long-term toxicity in rodents: in 90 day-studies one would need at least 10 animals per sex and per dose level, in 1-year studies at least 20 animals per sex and per dose level and in a  2-year study at least 50 animals per sex and per dose level. One should test at least three dose levels and also include adequate controls (a control group receiving the standard lab chow alone and one receiving the solvent in which the chemical is dissolved). It does not make a difference what sort of chemical you test."

7. What do the animal trials actually tell us? What ways are they analysed and what tests are done? (Scott Dwyer - @dwyer_s)

PS: "Animal trials tell us whether a compound can be toxic (health-threatening) in the short, mid or long term. We have animal trials in which the compounds are given once (acute toxicity study), for 28 days (subacute toxicity study), 90 days (subchronic toxicity study), 1 year (chronic toxicity study) or 2 years (carcinogenicity study). At the end of these trials blood and urine are collected and analyzed, animal weight and feed consumption are determined throughout the trials as a way of monitoring if everything is fine with the animals and at the end the animals are killed, all organs looked at macroscopically (looked at with the unaided eye) and microscopically to detect any lesions in them."

6. Will animal feeding studies ever be enough to prove GMOs are safe or are other kinds of tests needed? (Branwen Brockley)

WH: "It is never possible to provide absolute proof of safety, just to show that any risk is extremely unlikely. This is equally true of GM and non-GM ingredients. Tests other than feeding studies are needed and are used routinely. For example compositional analysis will show whether unexpected new proteins are present that might be toxic. The key thing to focus on is the exact modification made as this will inform the safety assessment in terms of which tests might be sensible."

5. What is the best animal to use as a model for humans in a feeding trial? (Eleanor Gilroy)

PS: "The very best animals to use would be pigs because they have similar metabolic capacities to humans. But they're not a practical choice as one needs large amounts of the compound to test (they eat alot) and if one wanted to look at long-term effects of a compound such as carcinogenic potential, it would take an awful long time to see any results. So in this respect rodents are a compromise as they are relatively small and one only needs two years to test carcinogenicity."

4. Are all crop plants (including non GM) regulated in the same way? Do all crops have to go through animal feeding studies to show they're safe? (Linsay Trerise)

JS: "No, only GM plants and derived products are regulated in a very strict way in Europe and have to go through a toxicological assessment including animal feeding studies."

3. Why do european regulations specify 90 day feeding trials?

PS: "To identify potential adverse effects and to address uncertainties regarding the potential toxicity of genetically modified food and feed. That's what the European Commission say themselves. The decision was made to be sure not to oversee possible toxic effects, but one could argue (and a number of critical voices do) that many toxic effects may not become evident within 90 days and that toxicity could be only revealed with longer studies such a 1 or 2 year feeding trial."

JS: "The new regulation is the outcome of a discussion process between the European Member States and the European Commission to incorporate the science-based EFSA Guidance for the risk assessment of GM food and feed into a legal text. Some elements including the obligatory 90-day feeding study were newly introduced by the Member States and the European Commission and are not in line with previously published EFSA opinions. According to the preamble of the regulation these elements were in part prompted by the views of some Member States and the desire to improve consumer confidence in GM food and feed."

"The current uncertainties in relation to the need and design of 90-day feeding studies are being addressed by the EU-funded research project GRACE (see The requirements regarding animal feeding trials in the context of GMO risk assessments will be reviewed in the light of the outcome of GRACE by the end of 2015 and other credible scientific knowledge which might be available at that time."

2. Is there actually any risk of eating GM crops before being tested on animals? Could GM create unexpected toxins in crops? (Linton Lahoud)

WH: "If GM crops are approved for human consumption then they will be safe to eat as extensive analysis will already have been carried out. In theory it is possible for GM to produce an unexpected toxin in a crop. However, other plant breeding technologies could also in theory produce unexpected toxins. There has been a vast amount of research looking for such unexpected consequences of GM and nothing has been found."

1. What is an animal feeding study? When were they first used to test crops? (Max Goldman)

JS: "In recent decades extensive experience has been built up from nutritional and safety testing in animals of novel foods, irradiated foods, fruits and vegetables. These experiences can also be relevant for the nutritional and safety testing of whole GM food and feed which have been conducted since the first experimental releases of GM plants in the nineties. Several studies have been carried out with feed derived from GM plants in target animal species like pigs, lactating dairy cows, sheep, broilers, and fish to assess their performance and their nutritive value. Even more feeding trials have been conducted for the safety and nutritional assessment of whole GM food and feed. GM foods derived from soybeans, maize, rice, potatoes, and tomatoes, for example, have been fed to rats or mice, and parameters such as feed consumption, body weight, organ weights, blood chemistry and histopathology have been measured."

"In order to test the potential subchronic toxicity (90-day trial), chronic toxicity (1-year trial) and carcinogenicity (2-year trial) of chemicals, four OECD Test Guidelines are available. However, these internationally accepted guidelines were designed and mainly used to test the toxicity of chemicals and not that of whole food or feed. The aim of the 90-days rodent (rats or mice) feeding study with whole GM food and feed is to assess potential unintended effects of toxicological and/or nutritional relevance and to establish whether the GM food and feed is as safe and nutritious as its traditional comparator."



Our Q&As answer the questions people put, which may mean that some parts of a subject are covered well and others not. If there is an issue that you think is not tackled, you are welcome to send a follow up question to our plant science panel