New study throws doubt on findings of safety in pesticide and GMO studies. Claire Robinson reports

Laboratory rodent feeds are highly contaminated with pesticides, toxic metals, PCBs, and GMOs, according to a new study soon to be published in the journal PLOS ONE.

The study casts doubt on claims of safety drawn from hundreds of thousands of animal feeding trials performed for regulatory approvals of pesticides and GMOs.

For the study, the team of Prof Gilles-Eric Séralini at the University of Caen in France analyzed the dried feed of laboratory animals sourced from 5 continents. These diets are commonly fed to rats used to test the safety of pesticides and GMOs. The study investigated 13 samples of rat feed for traces of 262 pesticides, 4 heavy metals, 17 dioxins and furans, 18 PCBs and 22 GMOs.

The researchers found that all the feeds contained significant concentrations of several of these products at levels likely to cause diseases by disrupting the endocrine and nervous system of the animals. Considering all the contaminants measured, these diets, when consumed over a long-term experimental period, would be considered by standard measurements to pose a very high hazard to health.

For example, residues of glyphosate, used on 80% of GMO crops and widely used to “dry down” non-GMO crops before harvest, were detected in 9 of the 13 diets. Eleven of the 13 diets contained GMOs that are grown with large amounts of Roundup.

This is a problem for public health because regulators use tests on animals fed on these diets to assess the safety of any one pesticide or GMO by looking at the difference between the exposed animal and the controls. If the treatment (exposed) and control groups are both eating an uncontrolled assortment of pesticides or GMOs, any actual toxic effect arising from the pesticide or GMO under test, unless the effect is massive in size, will be lost amid the “noise” caused by the jumble of potentially and known toxic substances.

The upshot can be that regulators conclude that the pesticide or GMO under test is safe on the grounds that no significant difference is found between exposed and control groups, when in fact both groups are exposed to such a wide variety of toxins that their effects have drowned out any toxic effect from the pesticide or GMO being investigated.

Treatment and control groups both fed uncontrolled GMOs

In the new study, eleven out of 13 of the diets were found to contain GMOs. This is of concern because these standard feeds are commonly used to test the safety of GMOs in the 90-day rodent feeding trials demanded by the EU regulatory system.

One of these diets, from the Purina company, was used in a 90-day rat feeding study by DuPont authors, which could be used to obtain regulatory approval of a GMO Roundup-tolerant canola variety. The study concluded that the GM canola was as safe as the non-GM canola, based on the lack of differences between the group fed the GM canola and the control group fed non-GM canola. Yet the new study found that the Purina feed contained around 12.8% GM soy and 35.6% GM maize – and was not even labelled as GM.

The new study also found that the feed contained residues of glyphosate and AMPA (the main metabolite or breakdown product of glyphosate). So although the control rats were not eating the GM canola under test, they were eating other GMOs with the same glyphosate-tolerant trait, as well as residues of the pesticide that the GMOs are grown with. In other words, the study did not compare rats fed a GM diet with rats fed a non-GM diet, but rats fed one type of GMO plus pesticides with rats fed similar GMOs plus pesticides.

Claims of safety based on such badly controlled studies are not worth the paper they are written on.

New findings challenge abuse of historical control data

The new findings also challenge the common practice by regulators to approve pesticides and other chemicals based on comparisons of the test animals not only with the control group within the experiment, as good scientific practice demands, but also with so-called “historical control data”.

Historical control data are collected from a wide variety of experiments, some decades-old, done in different conditions, in which animals are fed diets that will vary widely in levels and types of contaminants. Unsurprisingly, given these massive uncontrolled variables, the range of incidence of any one disease or toxic effect found in the historical control data will be huge.

For example, according to these data, 13–71% of control animals in an experiment would spontaneously or naturally present breast tumours and 26–93% would present pituitary tumours. Thus if the treated (exposed) animals present an increased incidence of tumours, if the increase falls somewhere within these very wide ranges, it could still be dismissed as “spontaneous” and not related to the substance being tested.

This wide range of “spontaneous” tumours also means that researchers have to use a large number of animals in an attempt to obtain statistical significance in the results of carcinogenicity tests, for example.

This is a matter of great public interest. In animal feeding experiments carried out for regulatory approvals of any one pesticide, industry and regulators routinely use historical control data to dismiss findings of an increase in a disease in the group of animals exposed to the pesticide over and above the incidence of the disease in the controls within the experiment.

They do this by pointing to the extremely wide range of disease incidence found in the historical control data and saying that as the increase in this disease found in exposed rats falls within that range, it can be considered “spontaneous” and can be dismissed. Thus the pesticide is wrongly judged safe – and public health is placed at risk.

Historical control data used to dismiss birth defects from glyphosate

Several years ago I came face to face with an example of the abuse of historical control data to “disappear” toxic effects from glyphosate. I co-authored a paper on the German regulatory authorities’ analysis of industry tests on glyphosate. We found that time and time again, the German authorities dismissed findings of birth defects in laboratory animals fed glyphosate on the basis of comparisons with unspecified and unpublished historical control data.

Since the number of birth defects in the animals fed glyphosate fell within the wide range of incidence found in the historical control data, the German authorities concluded that the birth defects were not related to the glyphosate treatment but were “spontaneous” – in other words, they could have arisen without exposure to glyphosate.

The new study shows this conclusion to be invalid. It suggests that there may be nothing “spontaneous” about tumours found in control animals. Far from being unexposed, they are in fact exposed to a large number of known and suspected carcinogens and tumour-promoters.

Eagle-eyed readers will recall that the Séralini team’s study that found a trend of increased tumours in rats fed GM NK603 maize and Roundup was dismissed by GMO proponents on the grounds that the number of tumours in exposed rats fell within the range of “spontaneous” tumours found in the historical control data on the supposedly tumour-prone Sprague-Dawley rat.

The new study’s findings suggest that such a judgment was ill-considered. In reality the NK603 maize study was a rare example of an experiment where the researchers controlled for pesticide and GMO content in the diet. In light of the new findings, the NK603 study appears to have been carefully carried out, while the studies routinely accepted as proof of GMO safety seem haphazard.

GMO proponents may wrongly exploit these results

Despite all the above, we must not allow GMO proponents and the European Food Safety Authority to exploit the new findings to argue for abandoning Europe’s currently mandatory requirement for animal feeding studies with GMOs on the grounds that it is too difficult to control the feed contamination content.

This is an outcome that GMO proponents have long pushed for. It would also please the US representatives in the TTIP trade deal negotiations, which aim to dismantle the EU approvals process for GMOs.

Instead the new results should push animal feed companies to clean up their act. The fact that no GMO contamination was found in the Italian animal feed samples shows it’s possible to make laboratory rodent diets without GMOs. Regarding pesticides, researchers (or the feed company) need to have to have the crops grown especially, with all chemical inputs carefully controlled at the farm level.

Animal welfare problem

Feeding lab animals uncontrolled toxins isn’t just a scientific problem: it’s an animal welfare issue.

Many animals in long-term experiments die before the planned sacrifice time (two years in the case of cancer studies). Typically the cause is kidney disease and tumours – diseases that are associated with the particular contaminants found in the animal feeds tested in the new analysis.

Researchers have to compensate for the premature deaths and ensure that enough animals are left alive at the end of the experiment to obtain statistical significance in the results. They do this by designing experiments using large numbers of animals. Purer diets would enable scientists to use fewer animals because there would be less statistical “noise” created by toxic contaminants.  

Society tolerates animal experiments if there is a perceived benefit in terms of protecting public health and the environment from the release of potentially toxic GMOs, pesticides and other chemicals. But in the case of experiments conducted with contaminated feeds, there is no payoff for the animals’ suffering in terms of generating findings that can help protect public health because the lack of control regarding the types and amounts of toxins mean no meaningful information can be got from the experiment.

The potential problems caused by contaminated diets stretch way beyond the laboratory. That’s because the same companies that make the laboratory rodent feed diets analyzed in the new experiment also make pet food. So it’s not only lab rats that are at risk from contaminated diets, but people’s dogs and hamsters.

The take-home message from this study is that feed companies have to improve the quality of their products. This will enhance the reliability of scientific experiments, improve the protection of public health, and create a better quality of life for lab animals and pets alike.


The new paper is published as:
Mesnage R, Defarge N, Rocque LM, Spiroux de Vendomois J, Seralini GE. Laboratory rodent diets contain toxic levels of environmental contaminants: implications for regulatory tests. PLoS ONE 2015: 10.1371/journal.pone. 0128429.