NOTE: The following are extracts from an important new report that shows in detail how the European Food Safety Authority (EFSA) used double standards to attack Seralini's paper. We recommend reading the report in full. It is only 14 pages in length including all the references and is very clearly written, and is accessible to the non-specialist. It also includes a useful table (page 11) directly comparing the different aspects of Seralini's study with other studies that EFSA has accepted without problem to support the approval of GM foods.
EXTRACTS: ...we believe that none of [the studies discussed in this briefing] describing chronic feeding studies with genetically engineered plants have met the standards being used by EFSA to criticize the research of Seralini et al., 2012. The comparison with the studies reviewed by Snell et al. (2011) shows that Seralini's research was conducted using comparatively higher scientific standards.
In conclusion, EFSA does not apply OECD standards to sub-chronic, 90-day feeding studies when these are prepared by industry and do not show health effects from consuming genetically engineered foods. In contrast, the OECD standards have been used by EFSA to attack the research of Seralini et al., 2012.
...The results should be taken seriously and used as a starting point for further investigations, as has been proposed by the French food safety and biotechnology authorities (ANSES 2012).
The European Food Safety Authority: Using double standards when assessing feeding studies
A Testbiotech background, 30 October 2012
Opinions of the authority show a bias in the scientific standards applied to risk assessments of genetically engineered plants
Recently published research on chronic (long term) animal feeding trials using genetically engineered maize (NK603) and the herbicide Roundup has been harshly criticised by the European Food Safety Authority (EFSA 2012). The research was led by French Professor Gilles-Eric Séralini of Caen University (Seralini et al. 2012) and although it was published in the peer-reviewed journal Food and Chemical Toxicology, the results were criticised by EFSA for not meeting specific scientific standards such as for example set out by the OECD (EFSA 2012a).
However, detailed analysis of former EFSA opinions shows that the authority has not taken a consistent approach when examining such scientific research. On a number of past occasions, EFSA has accepted without question the results from publications, on the risk assessment of genetically engineered plants, that are not in accordance with the scientific standards now being applied by EFSA to criticise the French study.
Unlike Seralini et al. (2012), these earlier studies did not conclude that there were any health impacts from eating genetically engineered plants. This inconsistency suggests that EFSA is 'picking and choosing' when to apply the scientific standards. There also is evidence that the food safety authorities of EU Member States are using similar double standards. For example, the Netherlands Food Safety Authority (NWVA, 2012) referred to a review paper by Snell et al. (2011) in order to refute the findings of Séralini et al. (2012). The review covered 12 chronic feeding studies using genetically engineered plants, and concluded that they did not show any health risks. But an analysis of the studies included in Snell's review shows that none of them met the same scientific standards that are now being applied by NWVA and EFSA to criticise Séralini et al. (2012).
In some respects, the standards used by Séralini et al. (2012) appear to be higher than those of the studies being used by EFSA and NWVA to refute his findings. Seralini's research seems to be the most comprehensive long term health study on genetically engineered plants to date. Also the French national food safety authorities (ANSES, 2012) – despite their criticism - have noted that the range of criteria examined was far wider than other long term studies. Furthermore, the German authorities (BfR, 2012) have pointed out that this research is the only long term study anywhere in the world to assess the health risks of the herbicide formula Roundup. The response to the findings of Seralini et al. should be new experimentation, rather than ad hoc refutations founded in assumption-based reasoning.
By failing to challenge the scientific standard of studies which do not show adverse health effects from genetically engineered crops, while at the same time attacking studies that indicate evidence of harm, European Union authorities such as EFSA are applying double standards and follow a biased approach. The authorities seem to be influenced by the presumption that genetically engineered plants should be regarded as safe and seem to be using the debate on scientific standards to defend their own opinions.
In view of the findings of Seralini et al (2012), the burden of proof should be shifted back to industry. Genetically engineered maize NK603 and the Herbicide Roundup maize cannot be regarded as being safe, so long as their safety is not proven by further investigations.
Further, in view of the debate arising from Séralini et al. (2012), the standards used in recent years for the risk assessment of genetically engineered plants and pesticides should be revised and reshaped, in order to achieve a higher level of protection for consumers and the environment. In addition, independent risk research should be promoted by EU research programs with a much higher priority...
[on EFSA's criticisms of Seralini's research]
...we consider EFSA's statement on this issue [the spontaneous occurrence of tumours in Sprague-Dawley rats, the strain used in Seralini's experiment] to be misleading, particularly the assertion that the spontaneous occurrence of tumours in these rats was "neither taken into account nor discussed in the Seralini et al. (2012) publication." Seralini et al demonstrate two different findings, both an increase in the number of tumours and an earlier onset of tumours in rats fed on the GE plant or the herbicide. Thus the researchers did not just measure the number of tumours over the lifetime of the rats. Further to support their conclusions, Seralini et al. (2012) compared their results to tumour rates in other published studies using this strain of rat. So they do appear to have been aware of the issue, and to have taken it into account.
EFSA's assessment of subchronic feeding studies
EFSA does not request sub-chronic or chronic toxicological feeding studies when assessing the health risks of genetically engineered plants. But in many cases, results from sub-chronic feeding studies are voluntarily provided by product developers.
Séralini et al. (2012) set out in their research to test the value of 90 day sub-chronic feeding trials. Thus they designed their experiment to conform with OECD guideline 408 for 90-day trials, but then extended the length of the test to examine further health impacts which became apparent towards the end of the 90 days period. The researchers used ten animals per group, which is in accordance with OECD Guidelines 408 for subchronic feeding studies.
We are not aware of sub-chronic toxicological feeding studies having been accepted by EFSA with fewer than 10 animals in each test group, as required by OECD (OECD Guideline 408), thus in regard of number of animals, the OECD criteria seem to be fulfilled. However the OECD Guidelines also request at least three dose levels. In other words, genetically engineered plant material should be included in the feed of test animals at low, medium or high concentrations. This allows the study to establish if there is a relationship between the dose and any observed health effect.
Seralini and colleagues did use three different dose levels in their study. In contrast, and despite OECD guidelines, EFSA routinely accepts 90 days studies with only two dose levels of genetically engineered plants in the animal feed. Indeed, when EFSA originally assessed the safety of NK603 maize, the 90 day feeding study submitted by Monsanto used only two dose levels of the genetically engineered maize (Hammond et al. 2004).
So 90-day feeding studies submitted to EFSA by product development companies do not routinely meet OECD guidelines, and yet the results are accepted by EFSA.
The design of sub-chronic feeding studies that have been accepted by EFSA have also not been in accordance with OECD standards in respect of the numbers of test groups used in the trials. In most cases, biotech companies use many more groups for comparison than is required by the OECD guidelines. The OECD Guideline 408 proposes only one group for comparison and a smaller additional ("satellite") group for ensuring quality standards, containing only five animals.
Monsanto's feed trials for NK603 maize (Hammond et al. 2004) used ten groups (per sex) but only two of these groups were fed with the genetically engineered maize NK603. All other groups, fed with various maize varieties (not just the isogenic line), were used for comparison and for additional references. Overall, 80 animals were fed with genetically engineered maize, but their data was compared against that from 320 animals from the reference and control groups. As a result, the statistical noise stemming from the additional reference groups threatens to cause a bias in the interpretation of the data, hiding relevant biological effects and significant differences.
90-day feeding studies, such as that by Hammond et al. (2004), follow standards set by industry. Although the OECD guidelines were not adhered to in these studies, the results were accepted by EFSA. Having accepted several industry-sponsored feeding studies using these standards, including those for genetically engineered maize MON810 and MON863, EFSA published its own guidance on 90-day feeding studies for whole food and feed (EFSA 2011). The standards in EFSA's guidance show little difference to those established by product developers which have a vested interest in the outcome of the risk assessment. In effect, the industry standards were adopted as the official standards of EFSA, with the explanation that OECD standard 408 was not adequate for this type of investigation.
So, the experimental designs used for 90 days feeding studies that are accepted by EFSA are not in compliance with the OECD standards. As a consequence, the findings and the methodology of these 90 day feeding studies have been a matter of controversial debate for years (see for example Spiroux et al., 2009).
In conclusion, EFSA does not apply OECD standards to sub-chronic, 90 day feeding studies when these are prepared by industry and do not show health effects from consuming genetically engineered foods. In contrast, the OECD standards have been used by EFSA to attack the research of Seralini et al., 2012.
On the basis of the studies discussed in this briefing, we believe that none of them describing chronic feeding studies with genetically engineered plants have met the standards being used by EFSA to criticize the research of Seralini et al., 2012. The comparison with the studies reviewed by Snell et al. (2011) shows that Seralini's research was conducted using comparatively higher scientific standards (see table 2).
In this context, it also should be recognised that the OECD Guidelines are not based on a process that can be considered independent from vested interests. For example, the International Life Sciences Institute (ILSI) is mentioned as a source in the OECD Guidelines. ILSI is an institution funded by the food and biotechnology industries. Beyond this, there is a wider issue of whether feeding studies, even if in compliance with the standards of the OECD, are adequate to assess health risks of genetically engineered plants. That is why according to the EU Commission (2012), the current standards will be reviewed within the next few years:
"The requirements regarding animal feeding trials in the context of GMO risk assessments should be reviewed in the light of the outcome of this project expected to be available by the end of 2015 at the latest."
In comparison to other studies, Seralini and colleagues used the most comprehensive criteria to assess health impacts, including a high number of samples and measurement of hormone levels (for more details see Seralini et al,. 2012, table 1). The French food safety authorities have also stated that the investigations in Seralini et al., ( 2012) covered a broader range of criteria than, for example, Sakamoto, et al. (2008). (See ANSES, 2012, table page 15 and also ENSSER, 2012).
Conclusion and recommendations
While there may be some problems with the methods used by Seralini et al. (2012) in their feeding trials, we believe their findings are still very important.
In comparison with previous feeding studies, mentioned by EFSA and NWVA, the research of Seralini et al. (2012) was conducted to higher scientific standards. The results should be taken seriously and used as a starting point for further investigations, as has been proposed by the French food safety and biotechnology authorities (ANSES 2012).
In contrast to their response to Seralini et al. (2012), when assessing applications for the approval of genetically engineered plants, European and national authorities often appear uncritical of the methods used by studies which do not show any adverse effects from the genetically engineered material. For example, the GMO Panel of the European Food Safety Authority has accepted such studies, even when they do not meet the OECD Guidelines for chronic toxicity or carcinogenicity studies.
From this evidence, it appears as if EFSA and some other EU authorities are taking a biased approach and applying differing standards during risk assessment. The outcome of their opinions seems to be influenced by a presumption in favour of the safety of genetically engineered plants.
This feeds a perception of selective evidence-gathering by EFSA, and that it favours the applicants who have a vested financial interest in the marketing of pesticides and genetically engineered plants.
After creating a history of certifying the safety of these products, it could be argued that EFSA and national authorities have a conflict of interest when it comes to reassessing their own conclusions. Indeed, their published statements on the work of Seralini et al. (2012) seem to be an attempt to refute any doubts about the safety of genetically engineered products and to defend previous opinions.
Based on our findings, we make the following recommendations:
The findings of Séralini et al. (2012) should be the subject of further experiment and investigation, rather than being dismissed. Given that consumers could be exposed to some of the relevant products each and every day, a high level of precaution is warranted.
In our view, EFSA has not earned a reputation of credibility for taking a final judgment on the research of Seralini et al. (2012). So we recommend that another scientific body be established on a ad hoc basis to deal with the assessment of this publication. That body should be assembled from appropriate experts free of financial conflicts of interest and independent of past contributions to EFSA decision making.
Genetically engineered maize, NK603 and the herbicide Roundup cannot be regarded as being safe, so long as their safety is not proven by further investigations. In view of the findings of Séralini et al (2012), the burden of proof should be shifted back to the product developer.
In view of the debate arising from the research of Seralini et al. (2012), the standards used in recent years for the risk assessment of genetically engineered plants and pesticides should be revised and reshaped, in order to achieve a higher level of protection for consumers and the environment.
Furthermore, independent risk research should be promoted by EU research programs with a much higher priority.