GRACE Stakeholder Consultation on Animal Feeding ... - GRACE FP7

GRACE Stakeholder Consultation on Animal Feeding Studies and In Vitro Studies in GMO Risk Assessment Draft Results of the 90-day Animal Feeding Studies and Progress of the In-Vitro, In-Silico and Analytical Studies with GM Maize FP7 Collaborative Project GRACE 311957

Report June, 2015

Authors: Gloria Adduci, Armin Spök supported by GRACE team members

Contact: [email protected]

This report must not be used as a resource for further research without prior permission of the corresponding author Armin Spök.

“GMO Risk Assessment and Communication of Evidence” (GRACE) is a Collaborative Project of the Seventh Framework Programme of the European Community for Research, Technological Development and Demonstration Activities. Grant agreement no: 311957 Project duration: 1st June 2012 – 30th November 2015 GRACE website: www.grace-fp7.eu

1

Table of Contents List of abbreviations ............................................................................................................................. 3 This Report ............................................................................................................................................ 4 Context ................................................................................................................................................... 4 Scope...................................................................................................................................................... 4 Approach ................................................................................................................................................ 4 Stakeholder Workshop......................................................................................................................... 6 Day 1: 19th May 2014 ........................................................................................................................... 6 Session 1: Welcome and Introduction ............................................................................................... 6 Session 2: 90-days Feeding Studies – Design and Analysis ........................................................ 9 Day 2: 20th May 2014 ......................................................................................................................... 16 Session 3: Alternative approaches .................................................................................................. 16 Session 4: Prospects and Networking............................................................................................. 19 Next Steps and closing remarks....................................................................................................... 24 Written Comments and Responses of the GRACE Team ........................................................... 24 Appendix 1: Agenda Appendix 2: List of Participants Appendix 3: Workshop Presentations Appendix 4: Comments Received in Writing Appendix 5: Organisations/Individuals Provided With Raw Data On Request

2

List of Abbreviations BCH CA CBI CSO DG RTD DG SANCO EASAC EC EFSA EU GM GMO GMP GRACE G-TwYST INRA MS OECD PreSto GMO ERA-Net RA SD SES US(A) USDA WP

Biosafety Clearing House Competent Authority Confidential Business Information Civil Society Organisation Directorate-General for Research and Innovation Directorate-General for Health and Consumer European Academies' Science Advisory Council European Commission European Food Safety Authority European Union Genetically Modified Genetically Modified Organism Genetically Modified Plant GMO Risk Assessment and Communication of Evidence GMP Two Year Safety Testing Institut national de la recherche agronomique EU Member State Organisation for Economic Co-operation and Development Preparatory Steps towards a GMO ERA-Net Risk Assessment Standard Deviation Standardised Effect size United States of America United States Department of Agriculture Work Package

3

This Report This Report summarises the presentations, comments and discussions at the GRACE stakeholder workshop held in Brussels on the 19th and 20th of May 2014 as well as the written comments received in response to the workshop and/or the workshop documents. It also includes the responses of the GRACE team to the written comments. Context In the course of the European Commission funded research project GRACE (GMO Risk Assessment and Communication of Evidence) animal feeding trials as well as analytical, in vitro and in silico studies are conducted with GM maize in order to test and further develop robust study designs and to evaluate comparatively their use for GMO risk assessment. In a first stakeholder consultation in December 2012 (workshop and written consultation) and in December 2013 (written consultation only) planning-stage issues relevant to these studies were discussed. The comments received were addressed when finalising the study plans. Stakeholder comments and GRACE-team responses are available at http://www.grace-fp7.eu/content/reportsstudy-plans-consultation-documents. The stakeholder consultation held in May 2014 focused on the first results of these studies. Scope The aim of this consultation was to: •

discuss the results of the 90-day studies with GM maize MON810

•

provide updates on the alternative and complementary studies to animal feeding trials (in vitro, in silico, and analytical studies), and to

•

provide information on follow-up and spin-off activities.

Approach In line with the first consultation a two-step procedure with a two-day workshop as the first and a time period for written comments as the second step was established. A first announcement was circulated by e-mail to 638 representatives from stakeholder organisations (competent authorities for GMOs in the Member States and at the EU level, industry, civil society organisations etc.) and to researchers active in this field. Those who responded received the workshop presentations and consultation documents summarising the results of the 90-day feeding studies before the workshop. Registered participants agreed to sign a non-disclosure agreement were also provided with the raw data of the 90-day feeding trial. Individuals interested but not able to attend were invited to send written comments. The workshop was held from the 19th to the 20th of May 2014 in Brussels. The meeting was audio recorded. After the meeting four weeks were given to provide comments in writing. The feed-back received in the course of the stakeholder consultation was summarised in a report drafted by the GRACE team and circulated to all participants to ensure that it accurately and fairly reflects their views and comments. Based on comments received the final version was prepared and published on the GRACE website. This report also includes the responses of the GRACE team to comments or questions received in writing and not sufficiently covered by answers or discussions at the workshop.

4

Participants This consultation was open to all stakeholders in GMO risk assessment Fifty-six representatives of stakeholder organisations (19 members of competent authorities, 12 members of industry, 3 representatives of CSOs, 22 members of academia) and 18 members of the GRACE team participated in the workshop (see Figure 1 and Appendix 2).

12

18

3

GRACE Team Members Academia/Research Competent Authority Civil Society Organization

19

Industry 22

Figure 1: Workshop participants divided by stakeholder category The meeting was hosted by GRACE partner PERSEUS and chaired by Gijs Kleter, Patrick Rüdelsheim, Joachim Schiemann, and Armin Spök. Reporting was done by GRACE partner IFZ. Written comments were received from 10 organisations or individuals (see Appendix 4). Access to raw data was provided on request to 12 organisations or individuals (see Appendix 5). Report structure The report is structured according to the agenda topics with a brief summary of the presentations followed by a summary of the comments, questions and discussions for each session. The latter ones are organised according to the topic and not necessarily reflect the chronology in which they were raised. A dedicated chapter holds all written comments and responses from GRACE team members. The comments are ordered according to topics. The original comments are provided in Appendix 4. The agenda, participant list, and slides of all presentations are provided in Appendix 1-3. Appendix 5 lists those organisations and individuals which/who ask for and were provided with access to raw data.

5

Stakeholder Workshop Day 1: 19th May 2014 Session 1: Welcome and Introduction Patrick Rüdelsheim (Perseus, Ghent, Belgium), organizer of the GRACE stakeholder workshop, welcomed the audience. He clearly explained the purpose of this workshop and described it as being an opportunity to present and review the results obtained so far in the 90-day animal feeding studies and alternative and complementary studies conducted in the course of the GRACE project. He invited participants to engage in the workshop debate through the Qs&As after each talk, in the discussions at the end of each session and by providing written comments during the workshop or via written comments in the following two weeks. He reminded that the presented data and distributed documents were confidential: GRACE researchers are in the process of publishing in a peer-reviewed scientific journal and any pre-publication of study results is forbidden. He further announced that all data including raw data, results and interpretations will be publicly available after the manuscript is accepted by the journal and will be published at the GRACE website. GRACE coordinator Joachim Schiemann (Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany) gave an Update on the GRACE project (ppt slides see Appendix 3) and underlined that GRACE aims to increase transparency and traceability of information dealing with potential risks and benefits associated with the use of genetically modified (GM) plants. His update on the project also included an overview of the allocation of the funding among the GRACE partners, the timeline of the project, the role of each work package in the GRACE project, the rationale for conducting animal feeding trials, the timeline of the 90-day and of the extended feeding studies (1 year), the sponsorship of public institutions, the project’s strategy towards transparency and stakeholder involvement and the revised project website. Finally, he presented established and future collaborations between GRACE and other EU and national research projects (PreSto GMO ERA-Net1, G-TwYST and GMO90+). Armin Spök (IFZ-Inter-University Research Centre for Technology, Work and Culture, Graz, Austria), WP coordinator in GRACE, presented the Stakeholder and User Involvement in GRACE (ppt slides see Appendix 3). He first highlighted the general approach and goals of GRACE involvement activities: to ensure involvement of stakeholders and professional users in key steps of the project in order to inform and fine-tune the research process and to enhance relevance of the project from a broader societal perspective. He then explained the different steps and activities of GRACE stakeholder engagement and provided an update on the GRACE planning stage consultation. The workshop program and timelines for following steps in the consultation were further detailed. Ralf Wilhelm, (Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany) focused on the Implementation of the feeding studies, requirements and constraints (ppt slides see Appendix 3). He gave an overview of the budget and of the amount dedicated to 90-day feeding studies, extended feeding studies, additional test variables, compositional analysis, plant omics and in vitro test. The update also explained resource constraints (e.g. availability of GM plants, additional costs and time limits).

1

http://www.presto-gmo-era-net.eu/ 6

Discussion Rationale for and design of the 90-day feeding studies 1.

Clarification was sought from a member of the academia/research community about the rationale for feeding studies: if 90-day feeding studies really have an added value, also considering that they are mandatory for any new event; and to which extent it was decided that and if there is a need to sacrifice more test animals given all the data already available in publications.

In line with Commission Implementing Regulation (EU) No 503/2013, the performance of 90-day feeding studies is now mandatory for all single GM events. However, the EC shall review the requirement to perform 90-day feeding studies in rodents with whole GM food/feed on the basis of new scientific information, which includes the results of the GRACE project. Regarding the second part of the question, the added value was highlighted: to provide data from a research project that is financed by public money generated with a high level of transparency and with involvement of stakeholders. The final GRACE stakeholder workshop will include data from the feeding studies, the in vitro methods, the omics as well as data obtained from cooperation with other projects and will provide a good basis for scientific and evidence based decisions. 2.

A member of the industry asked about the scientific reasons for the chronic feeding study and the decision to start them before having the results of the 90-day feeding studies.

A GRACE team member clarified that in the FP7 call an extended feeding study was requested besides the 90-day feeding studies and in the negotiation phase the EC specified that this task needs to be fulfilled. GRACE started by conducting literature reviews, developing the design and this was extensively discussed with stakeholders during the first GRACE stakeholder workshop [held on 3-4 Dec 2012 in Vienna]. In this workshop it was repetitively asked by stakeholders to perform this 1-year extended study. For these reasons it was decided to perform a 1-year extended study. Waiting for the results of the 90-day feeding studies before performing the 1-year chronic feeding study would have been a good idea but time constraints to produce the feeding material put the GRACE team under pressure to decide upfront which kind of extended study to should be performed. 3.

Clarification on the design of the 90-day feeding studies was asked by a member of the industry, specifically on the differences between the approved design (claimed to be designed according to scientific literature) and the one suggested in the EFSA guidance.

A GRACE team member explained that it was decided to include in the study two different MON810 varieties (one Monsanto, one Pioneer variety) containing the MON810 event in their different genetic backgrounds, which are compared with their respective non-GM control varieties (instead of a single GM variety usually tested in published studies and also described in EFSA guidance). This is done in order to observe what is the possible influence, if any, of plant materials with different genetic background, used in feeding trials with whole feeds and whether differences between a group fed a MON810 variety and that fed its counterpart are also (consistently) observed between the groups fed the other MON810 variety and tits near-isogenic counterpart. Four additional non-GM conventional varieties are used in order to produce historical data that are directly linked to animal experiments with whole foods containing diets, which were not available in Bratislava for animals fed this particular type of diet and to get additional information on possible biological variability within the rat strain used on the basis of different whole foods. 4.

A member of academia/research community added that it seems there is no clear explanation on how the choices on feeding studies were made other than that it was already done in this way in previous studies. It seems of limited value and he wonders what GRACE could add to the already present knowledge.

7

5.

An industry representative asked about group sizes: the traditional size of 10-12 animals per group is already powerful enough (industrial review of what effect sizes is toxicologically meaningful); it was added that standardized effect size does not equal toxicological relevance. Industry suggests not using more animals than necessary; hence, no need for 16 animals.

A GRACE team member commented that the decision of 16 animals was made based on the first stakeholder workshop [held on 3-4 Dec 2012 in Vienna]. Two possibilities were proposed at this workshop: 10 animals or 16 animals but experts in statistics strongly suggested 16 animals. Another GRACE team member added that there indeed is a distinction between statistical and toxicological relevance, yet they nicely complement each other; group size is also of socio-economic importance as recommendation; no exact thresholds for some parameters was possible; the two studies showed different results. He added that industry has data showing that 10/12 animals per group are enough, but such data are not available for the lab in Bratislava, so that a cautious approach needs to be followed. Again another GRACE team member commented that finding the right size of the animals groups was an issue of attaining the statistical power deemed sufficient for the experiment, as explained in the EFSA Scientific Committee's guidance document. 6.

A representative of industry asked about the study design: why have different designs have not been compared.

7.

A representative of academia/research community commented that the power of feeding studies is overestimated, and that alternative methods are more sensitive. He added that the question if 90 days are sufficient [compared to longer time periods] is always open. Chronic studies are requested by the EC mainly for political reasons, not scientific reasons.

8.

An EFSA representative explained that the EFSA guidance requires the presence of appropriate historical control and background data.

It was clarified by a GRACE team member that the issue of including additional conventional varieties was already debated in the first stakeholder workshop [held on 3-4 Dec 2012 in Vienna], where a representative of EFSA claimed that this is not in line with the EFSA guidance. However, due to the fact that the colleagues in Bratislava who performed the feeding studies had no previous experience with GM plants and therefore no respective historical data generated in similar types of experiments, it was deemed prudent to produce them and to generate additional data on biological variability within the rat strain. This can be complemented with the data provided by the company that provides the rats for the experiments. Feeding studies with potatoes 9.

Clarification was sought from a member of academia/research community about the decision of the USDA to retreat from the cooperation with GRACE and not to provide GM potatoes.

A GRACE team member clarified that USDA provided scarce explanation for the decision, mainly that this project did not fit into their scope for research. 10. A participant from the academia/research asked if the decision to drop feeding studies with potatoes was definitive, since it would be interesting to compare the results with the two maize MON810 used in the 90-day feeding studies. 11. A participant commented that it would be a mistake not to conduct the potato feeding study because he doesn’t think that there are nutritional issues when feeding potatoes. He asked for clarification on how the potatoes would be fed to rats, since it makes no sense to feed them raw, but they should be cooked in the way humans consume potatoes.

8

In responses to these questions/comments GRACE team members clarified that the decision to cancel the feeding studies with potatoes had not been taken yet and that GRACE partners wanted to raise this topic at the stakeholder workshop, asking for suggestions on which kind of study would be best to perform with potatoes. The original plan was to perform a study with potatoes obtained from USDA. The USDA is no longer partner so the only options are the cisgene potatoes which are also available in limited amounts only. On top of this, budgetary constraints need to be addressed. In response to the discussions at the first stakeholder workshop, additional analyses were performed, which were not considered in the original budget (no additional funding will be provided by the EC). The discussion in this workshop will be considered together with the GRACE Project Officer in light of the limited budgetary flexibility. If new analyses on potatoes need to be included, it is likely that some of the analyses on maize need to be dropped. Alternative options are to perform in vitro studies or reviewing the published studies of feeding trial(s) with hamsters (90-day feeding studies with potatoes), which could give some useful results (hamsters are more sensitive to glycoalkaloids) or a desktop research study reviewing previously generated data on laboratory animals fed similar GM products; such existing studies could be considered in the overall discussions of the project results and recommendations. It was added by a GRACE team member that time constraints are very important due to the duration of the studies and the end of the GRACE project, foreseen at the end of November 2015. An extension of the project will be applied for, but it is not sure that the Commission will grant it; moreover, results need to be available in time to inform the Commission for the review of Implementing Regulation (EU) No 503/2013. 12. A representative of academia/research raised the issue of cisgenic potatoes. A GRACE team member clarified that the initial focus was on potatoes with altered glycoalkaloid profiles in order to consider the sensitivity of 90-day feeding trails. Trials using cisgenic potatoes cannot address the issue accordingly. 13. Among the comments received from stakeholders, a representative of the academia/research community mentioned that it would be essential for GRACE including at least one or two test groups with some toxic substances to look for the sensitivity on the test (e.g. spiking). 14. A member of the academia/research community made a remark on purpose and hypothesis of feeding trials, specifically on potatoes. He suggested that there are far better models for glycoalkaloids. Lots of data are available on this topic, but still there are many uncertainties among political experts; in their opinion, this fact might be due to the fact that no clear hypothesis is made before trials. GRACE may be unable to provide scientific arguments. He stressed the importance of having a clear idea of the reasons behind these experiments and a precise hypothesis from the beginning and not afterwards.

Session 2: 90-days Feeding Studies – Design and Analysis Pablo Steinberg (University of Veterinary Medicine, Hannover, Germany), member of the GRACE Task Force on Feeding Studies, gave an overview of the General design and analysis for the 90-day feeding studies (ppt slides see Appendix 3). Starting with the objectives of the two 90-day feeding trials conducted by GRACE he explained in detail the study design including sample size calculation, the experimental unit, how the rat feeding studies were performed and which statistical tests were applied.

Pere Puigdomènech’s presentation (Center for Research in Agricultural Genomics, CRAG, Barcelona, Spain) on Maize Cultivation (ppt slides see Appendix 3) gave an overview of the MON810 maize 9

production and commercialization in Spain and specifically in Catalonia. Referring to the maize material produced for GRACE, he showed the locations of the fields and explained how all agronomic, morphologic, and health parameters for the maize were checked, and preparation of maize material for the feeding studies and compositional analysis and in-vitro studies within Work Package 2, including omics.

Gijs Kleter (RIKILT-Institute of Food Safety, Wageningen University and Research Centre, Wageningen The Nederlands), WP leader in GRACE, described the Analysis of Plant Material and Diets (ppt slides see Appendix 3). Referring to the maize material produced for GRACE, he provided some background on the material starting from its production in Spain and how the diets were prepared. He showed which analyses were performed and the results both of maize and of the whole diet (macronutrients, micronutrients, anti-nutrients, as well as the presence of DNA and protein from MON810 and other GMOs, microbiology and a range of possible contaminants).

Dagmar Zeljenková (Department of Toxicology, Slovak Medical University, Bratislava, Slovakia), WP leader in GRACE, described the Test facility, periodic observations and necropsy (ppt slides see Appendix 3). After introducing the facility, she showed the periodic observations and necropsy results from study A (Monsanto MON810) and study B (Pioneer MON810) and gave an overview of the findings and conclusions with essentially no findings of effects related to GMO.

Jana Tulinska (Department of Toxicology, Slovak Medical University, Bratislava, Slovakia) reported on Haematology and biochemistry (ppt slides see Appendix 3). She showed the haematological and biochemical results for both studies and concluded that, in one of the two studies with Monsanto MON810, only blood sodium changed levels were altered in both genders. For other parameters being significantly different to the values of the control group no dose-dependency could be identified. Concerning the study with Pioneer MON810, she concluded that there was no dose- dependency for parameters showing significant differences in both the 11 and 33% GMO groups.

Pablo Steinberg (University of Veterinary Medicine, Hannover, Germany) presented the Histopathological results of the 90-days feeding studies (ppt slides see Appendix 3). He described tissue sampling and formalin fixation. Details regarding the findings in both studies were given.

Discussion Animal diets and dosing 15. A representative of a CA asked how it is possible to determine how much feed each rat consumed, if two animals were kept in each cage. A GRACE team member replied that weekly measures included total feed consumption per cage and the weight of each animal. Two animals per cage are considered one “experimental unit”, as were specifically recommended by EFSA on the basis of animal welfare considerations. Historically, the parameters were for each animal were monitored, but in the two 90-day feeding trials the cage effect also was considered – all data are recorded per cage and per animal. In the manuscript to be sent to the journal, the data per cage will be provided. The data for each individual animal will be made available through the database CADIMA. 16. A representative of CA asked about the rationale for choosing 33% as highest dose, since some studies suggest even a higher dose?

10

A GRACE team member explained that 33% was recommended in the literature and therefore chosen. A GRACE team member added that the possibility of a higher dose was extensively discussed in the first stakeholder workshop and mentioned in written comments: many comments were made on the possibility that higher doses could create a nutritional imbalance; therefore, 33% was chosen as highest value. 17. One participant asked if other relevant differences between the GM maize and isogenic one other than the included Bt toxins were identified. A GRACE team member answered that, as shown in the workshop documents, some differences were observed, but these did not affect the outcome of the feeding trials. 18. A representative of academia/research community requested clarification on the standardised effect size (SES) choice of 1.0 standard deviations (SD) as the biological relevance limit. Moreover, he added that from a nutritional point of view it is possible to include a much higher dose, such e.g. up to 60%. He invited the consortium to consider the possibility of choosing only one dose, the highest possible one. GRACE can develop its own protocol and does not necessary need to accept values. A GRACE team member answered that the 1 SD was chosen due to a priori lacking historical data in the lab. A higher threshold value for the SES analyses would include data points that are spread out over a large range of values; this in turn was considered to be a less rigorous interpretation of the approach to evaluate the relevance of differences between two groups in the GRACE 90-day feeding trials. As to the possibility of including only one extremely high dose, the danger of running into nutritional imbalance and of not being able to determine if a difference is dose-dependent has to be considered. 19. A representative of CA sought clarification for the choice of 11% as low dose and on why haematological and clinical investigations were done one week before necropsy. Several GRACE team members responded by explaining that an 11% GMO dose was chosen because it represents a human-relevant dose. Concentrations of 11% and 33% of GM maize in the feed were suggested by test guidelines and recommendations from OECD and EFSA. Moreover, this study design had been confirmed in the first stakeholder workshop. Haematological and clinical analyses were performed one week before necropsy for technical reasons: 40 animals per day required too many people at the same time for performing full necropsy as well as haematological and clinical analyses. 20. A representative of academia/research community commented that the model used is rather insensitive and suggested to increase the sensitivity with a different dose. His concern is that a study with no clear hypothesis would not allow seeing what is really relevant. Whole GM food/feed studies showed no dose-dependency so far and this is due to the fact that the system is not sensitive enough, therefore he suggested using the highest dosage possible and evaluating if something is found under these conditions. A GRACE team member responded that using a single very high dose alone would not allow to determine whether an effect is dose- dependent and in terms of risk assessment probably is not the best choice. 21. A representative of CA commented that high content of protein in the diet can mask toxicological effects. Considering the used rats, the requirement of protein is 10% protein. If the protein content would be decreased the experiment would be more sensitive. A GRACE team member responded that the influence of the protein level on the toxicity outcome was not taken into consideration when planning the studies. In this respect, GRACE is making use of experience gained over years in the field of chemical toxicity testing, but so far it is not clear which will be the conclusions of the project when testing the GM plants.

11

A GRACE team member added that in the GRACE experiments feeding trials, two maize varieties with the same GM event were included, and if any effect is seen in both groups, it is more likely to be GMOrelated. Moreover, the point of a too high level of protein in diets was raised in the first stakeholder workshop. The issue was therefore discussed with nutritionists, and based on these discussions it was decided that the composition of the diet should not be changed. A GRACE team member commented that based on the results of the two 90-day feeding studies alone it is not possible to have elaborate definitive conclusive conclusions/recommendations for the future GMO assessment. At the end of the project, GRACE will combine all results from in vivo (90-day and 1year feeding studies), in vitro, in silico studies, and analytical studies and also consider results obtained from other related projects and might be able to draw interesting conclusions. When designing the two feeding studies, many nutritionists were consulted. Moreover, GRACE team members preferred to avoid a too high content of maize in the diet that could jeopardize the reliability of the data and the validity of the study. 22. A CA representative asked about the soybean lectins present in diets, given that the customdiet-producing facility used mild conditions of diet preparation. A GRACE team member explained that lectins were below quantifiable levels.

Relevance of traces of other GM events 23. A representative of academia/research community asked if there was a chance of crosscontamination between fields. A GRACE team member explained that there was no chance of cross-contamination: because of the cultivation design: the different fields are very isolated one from the other. 24. A representative of industry pointed out that based on in the data distributed before the workshop, there was evidence of traces of GM events and wondered if this might have happened in the fields or during further analyses. A GRACE team member clarified that very low levels of contamination were observed, and that these might have occurred already in the seed used for sawing or in the post-harvest stages. During the preparation of feed, there was a strict sequential order to first treat first non-GM and then GM seeds, therefore contamination was unlikely at that stage. The source of the contamination, however, was not clearly identified. 25. A representative of industry asked if the same batches of ingredients were used to prepare the diet. In that case, the source of contamination can be found in each diet. He also asked if there is any idea regarding the source of contamination. He then asked how was decided that certain levels of contaminants were considered acceptable for the mixture of the diet if in conventional varieties were detected low levels of contaminations. A GRACE team member answered that a single batch was used for each variety. Contaminations were measured at very low levels and were also difficult to be detected; for MON810, consistency between the maize and the diet was observed. It is difficult to understand where the contamination might have occurred, as the animal food/feed-producing facility took care to avoid cross-contaminations, the equipment was washed and first non-GM and then GM material was processed. Other GMOs such as NK603 were also detected at very low levels. It is considered that for these GMOs, most likely the contamination is from the sowing seed itself. There was no cut-off level set for contaminations. 26. A representative of CA asked if this GM contamination is a reason for concerns. A GRACE team member answered that very low levels of contaminations are unavoidable, and the methods used for detection were sensitive and validated. Moreover, there is still 98% of non-GM (in the 12

particular case of 2% MON810 measured in one of the conventional, commercial non-GM varieties used for the historic background, other maize varieties showing lower levels, if at all), so that a comparison between GM – and a commercial line would still be able to discern the impact of the particular GM trait, if any. 27. A representative of CSOs asked at what stage(s) the Bt content was measured and at which stages were the samples milled. A GRACE team member explained that the maize was milled at the feed-producing facility, while diet pellets were milled at RIKILT (to which samples had been sent by the feed producer). Care was taken to prevent overheating during diet preparation by choosing a mill sieve size (1 mm) that was not too small. The Cry1ab protein was measured both in the milled maize and in the diet. GRACE colleagues in France received additional samples of kernel coming directly from Spain for in vitro measurements. All the samples were coded, so that investigators did not know what was analysed (GM or control). 28. A representative of industry asked if multiple varieties with the same event would help to distinguish event- from germplasm-specific effects. A GRACE team member answered that the two MON810 varieties were chosen, since the Spanish partner already had experience and background with those varieties. Yet nevertheless, two is rather a limited number of varieties. At the moment, effects observed do not appear to be event-related.

Statistics 29. A representative of academia/research community commented that GRACE should also look at the methodology. Many statistical models for statistics and methods are available, so that GRACE should consider them and propose a more harmonized model. A GRACE team member commented that analysing the results of the feeding studies only in terms of statistical differences is not the correct way to proceed. The data should be discussed by taking into account the statistical relevance as well as the toxicological relevance of the differences observed. The SES analyses showed that statistics can be used to interpret the data, but these data will also have to be interpreted from the toxicological point of view. A GRACE expert added that having an SES of 1 SD or 2 SD is an arbitrary decision on the interpretation of the biological relevance of differences; the use of an SES itself grants a standardized approach. Concerning statistical methods, EFSA recommended an analysis following this approach and the GRACE consortium adopted these statistical approaches. 30. A representative of CA asked if - looking at the data obtained so far and at the statistics and biological differences - GRACE members still consider the choice of an SES of 1 SD appropriate. This approach has not been used very often. [GRACE team response: For these trials 1SD was chosen for not having a better guess. In fact it was just frankly following the example from the EFSA guideline. Yet the threshold value for relevance should ideally be set based on a detailed knowledge on variability and observed relevant effects in a comparable test system.] A GRACE expert added that this method is not replacing the ones used before but complementing them. It gives not only a ‘yes or no’ answer but also a size of what is observed. 31. A representative of academia/research community asked how the findings of haematology and biochemistry were considered in terms of statistical differences.

13

A GRACE team member answered that all animals were tested; historical background data are only available from literature. Sodium levels consistently increased, which could not be related to analytical problems. Outlier analysis identified outliers for sodium, but these values were not left out of the statistical analysis since reasons for a causal justification of exclusion were not identified.

Haematology, biochemistry and histopathology 32. A participant asked why histopathological parameters were not included in the statistical analysis. A GRACE team member explained that the histopathology results were not included in the statistical analyses due to scarce the number of findings. 33. A CA representative commented that differences in organs can relate to difficulties in sampling, (particularly referring to the pancreas which in the rats is a widespread organ in abdominal cavity) and asked which protocol was used; also asked clarifications about the relative organ weights calculated. Moreover, gastrointestinal organs weights should have been checked. A GRACE team member answered that no effects were observed in the gastrointestinal organs. A GRACE team member added that the recording of the pancreas weight is unusual and not included in the OECD Test Guideline, but was requested in the first stakeholder consultation. 34. A participant asked if there was an impact due to animal’s last meal. A GRACE team member answered that animals had been without food for 18 hours before being sacrificed. 35. A participant asked if all tests were performed under GLP conditions and why abnormal changes were seen in younger animals but not in the older ones. A representative from industry asked if sampling of animals was fully randomized. A GRACE team member answered that all haematological, histological and biochemical analyses were performed under GLP and that all 90-day studies were under GLP conditions. Results were not abnormal as evidenced compared with Harlan’s historical data. Sampling of animal cages was not done in a randomized way. 36. A member of academia/research community commented that biological interpretations are important and there is a general lack of ability to interpret data. 37. A CSO representative added that most differences appear to have occurred in the GMO groups and this would be interesting to be discussed. 38. A representative of CA commented that metabolic differences were observed between male and female animals. A GRACE team member responded that male and female have in fact different metabolisms and only cases where differences were observed were shown in the power point slides (for e.g. hormonal and metabolic differences). 39. A representative of CA commented that it would be interesting to see also histopathological results for 11% MON810 maize and not only for 33%. A GRACE team member answered that a histopathological analysis on samples of animals fed 11% GM maize was not performed. This would have been done if a considerable number of lesions were observed with the 33% GM diets, but this was not the case.

14

40. A member of academia/research community asked about the possibility of using also other strains of rats and if the use of Sprague Dawley is discouraged because it shows more changes than Wistar Han rats. A GRACE team member explained that based on the outcome of the two GRACE 90-day feeding trials the use of Wistar Han rats is highly recommended. 41. A member of academia/research commented that in his opinion there is no problem with the strain of rats used. 42. An industry representative commented that it is not normal to see neoplastic lesions in 90-day feeding studies and asked if the rat strains chosen for the feeding trials might be appropriate. Another issue is the tumours found. A GRACE team member confirmed that it is unusual to find pre-neoplastic lesions in rats, but the extremely low incidence of such lesions (one lesion in one out of 16 control animals) does not disqualify the rat strain chosen for the feeding trials. 43. A representative of CA asked if the study was reviewed by a second independent histopathologist. A GRACE team member clarified that it was peer reviewed by another an independent histopathologist. 44. A member of academia/research community asked if the consortium was able to look at data from Monsanto or Pioneer on histopathology. It was answered that the GRACE consortium did not have access to these data. 45. A member of academia/research community commented that findings of biological relevance could either be related to the presence of Cry1ab or something else. He asked how to differentiate these effects. A GRACE team member responded that the findings in the 90-day feeding studies justify not to assume being event-related. Results from the 1-year feeding trial might clarify 46. A representative of industry commented that if there is a size-effect correlation, then the big question remains “how many replicas are needed to show a conclusive effect?” Moreover, after the 1-year study we are likely to have the same issues/questions as now after the 90-day studies. A GRACE team member answered that this may very well be the case, but we still need the results of the 1-year study to assess the relevance of the results obtained in the two 90-day feeding trials. 47. A CSO representative observed that it seems that immunologic studies are not a topic for the 1-year study. A GRACE team member answered that immunological studies are included in the 90-day studies but not in the 1-year study. 48. A representative from CSOs/NGOs added that most differences appear to have occurred in the GMO groups and this would be interesting to discuss. 49. A member of academia/research community commented that, in order to perform a 90-day feeding trial to test for results on toxicity, a good hypothesis is necessary. A GRACE team member answered that a discussion on this topic also took place also at the first stakeholder workshop. Glycoalkaloid potato had a clear focus on toxins. Maize MON810 was chosen for other reasons (e.g. possibility to grow in Europe; commercialized GMO).

15

Other questions and comments 50. A member of academia/research community commented that biological interpretations are important and that there is a general lack of ability to interpret data. 51. A GRACE team member (not involved in the studies presented in the workshop) asked if looking at the data presented something can be already concluded on the outcome for risk assessment of GMOs. The costs of feeding trials should also be considered when assessing their value and developing recommendations (e.g. cost-benefit analyses, appropriate number of animals). A GRACE team member responded that it is too early to consider these aspects. In this context, the in vitro studies are of great importance. The GRACE final conclusions will be based on all results to be obtained from the in vivo, in vitro and in silico studies. 52. A member of the academia/research community commented that, due to the many changes between the draft study plans circulated for stakeholder consultations it would have been interesting for stakeholders to be more involved and informed about the changes that were included in the final study plans. A GRACE team member clarified that the first draft plan of the 90-day feeding trials has been published, along with all the amendments to that draft and also changes to the initial protocols. All these documents are available on the GRACE website.

Day 2: 20th May 2014 Session 3: Alternative approaches Ralf Einspanier (Freie Universität, Berlin, Germany), WP coordinator in GRACE, presented an update on the In Vitro Studies conducted in the course of the project (ppt slides see Appendix 3). He clarified the objectives of the study (omics, different type of samples, such as tissues or in vitro cultured cells), indicating the in vitro and cell culture systems as main focus of the topic. Tissue analyses (transcriptome and liver protein), in vitro exposure, cell culture data and dissemination were presented.

Esther Kok (RIKILT-Institute of Food Safety, Wageningen University and Research Centre, Wageningen, The Netherlands) introduced the Omics studies on plant material (ppt slides see Appendix 3). After an introduction on the background on comparative analysis of plants as one of the central items (focusing on both intended and unintended effects), she talked about the compositional analyses, the omics approaches (transciptomics, metabolomics, NGS sequencing) reporting some examples such as omics profiles of potatoes and maize and the statistics employed to analyze the outcomes.

Karine Adel-Patient (Unité INRA d'Immuno-Allergie Alimentaire, Gif-sur-Yvette, France) presented the Subchronic toxicity and longitudinal metabolomics studies (ppt slides see Appendix 3). She introduced the aims and methods of the studies performed, focusing then on the metabolomic studies (principles, integration and interpretation of the data, statistics), on subchronic toxicity and finally on the immunological status of Cry1Ab and of maize.

Discussion 16

Method validation 53. A member of the academia/research community asked how the system will be validated (e.g. candidate genes, pathways). A GRACE team member explained that selected candidate gene (gRT-PCR) as well as complete transcriptomics (RNAseq) analyses will be performed (these methods can be very sensitive). 54. A stakeholder from academia/research community considered that the omics techniques are very promising but they require validation for their application in risk assessment and so far little progress is made on this issue. A GRACE team member answered that in the frame of GRACE the protocols are being fully standardized and it has to be considered that animal feeding studies are not a validated process as well. So far the protocol for the omics techniques is not validated yet; it requires high quality data that can be reproduced. 55. Referring to validation, a representative of EC suggested keeping a very good and extensive documentation of all the procedures, also for the in vitro studies, allowing for reproducibility of the methods. A GRACE team member clarified that this is in complete agreement with the key objective of GRACE to provide a high degree of transparency. Reports on previous workshop as well as Qs&As are available on the GRACE website and the same will be done for this workshop. Everything is documented, subject to public scrutiny, criticisms and discussion with stakeholders.

In vitro methods 56. Stakeholders from academia/research and industry pointed out that salt level in the system should be evaluated (effects on the cells due to the salt were registered) and that composition of maize extracts and impact of solvents should be taken into consideration. A GRACE team member answered that media conditions need to be fully controlled (effects on the cells are taken into consideration); moreover the method used is based on water extraction, which gives fewer problems than organ solvents. A protocol for aqueous maize kernel extraction has been established. 57. A representative of industry asked about the number of replicates and commented that the staring material should be well characterized. A GRACE team member clarified that replicas on many samples are performed, due also to the need of many control reactions to identify false positive or false negative. 58. A stakeholder from academia/research community asked about the possibility to include intestinal juices. A GRACE team member clarified that this is performed. Specifically pre-digestion protocols are used in close cooperation within the GRACE consortium (INRA). 59. A stakeholder from CSOs/NGOs asked if the update from the OECD guidelines on genotoxicity will be taken into account. A GRACE team member answered that in theory yes, but it is not easy, since the levels of toxicity detected are very low. Moreover, the main focus of these investigations is not on toxicity but on minor effects. 60. A stakeholder from academia/research asked for a clarification on the minor effects, considering that they are measured with great sensitivity and variability among individuals, specifically how this is accounted for, also from an epidemiological view. 17

A GRACE team member answered that with artificial systems the approach is to test minor effects but there is the need to compare results with data from other approaches (in vivo, in silico). Once effects are revealed, it is necessary to go back to the biological system to check what they mean. Too sensitive methods could lead to larger numbers of false positives and this has to be considered. GRACE is trying to investigate complex systems using also new techniques that can give huge amount of info that need to be finally reconsidered and evaluated. 61. Representatives from academia/research asked to what extent can in vitro systems contribute, supplement or abolish in vivo systems, given the fact that different GRACE team members seem to have different views. It was noted that the value of feeding trials is very limited and these methods could be performed before the feeding studies and point to that if something is observed. In vitro methods represent a good alternative and need background and validation. GRACE team members specified that so far, results available are only partial due to the complexity of the systems that are being investigated (partial but still very promising) and there will be an integration of all results at the end of the project. It cannot be promised that these methods will substitute feeding studies but the comparison of the information obtained on the basis of animal feeding trials with the information obtained with alternative methods is a central and required element in the project. It is likely that these methods can be used to identify unintended effects. 62. A representative from academia/research community commented that a stepwise approach is already followed in GMO risk assessment. The quality of in vitro analysis of whole GM food safety is indeed a unique topic. It would be useful if GRACE could come up with recommendations, specifically on the in vitro studies. A GRACE team member answered that in vitro studies need to be adapted and validated but GRACE is working in that direction and might be able to give recommendations for biomarkers on sub-chronic toxicity.

Omics on plant materials 63. A stakeholder from academia/research suggested to consider unintended changes instead of unintended effects and asked about the purpose if those tests, if they are a replacement for another test, how much they cost and how fast they can give results. A GRACE team member believed that this test in the long run will replace compositional analyses. Before doing so, these methods need to be validated. Referring to costs, the costs associated with omics approaches should be compared to the costs of animal feeding trials and targeted compositional analyses. With respect to the costs of transcriptomics on the basis of Next Generation Sequencing, they are decreasing rapidly. 64. Representatives of industry asked if the whole analysis is done twice in same cases; it was also considered that, if the current method for analysing metabolomics is only 10% effective, the ability to make interpretations will be affected. A GRACE team member clarified that the analysis is repeated with two samples from the same batch. Referring to method and interpretation, it was clarified that so far transcriptomics gives the most conclusive results and maybe later on combination with metabolomics or proteomics can be possible. 65. A member of the academia/research community, referring to the one class model, asked how profiles out of the safe class should be interpreted. A GRACE team member clarified that it is necessary to include the right control. Values outside the safe class are not necessarily unsafe, but will need to be further looked into.

18

Longitudinal Metabolomic Studies 66. A stakeholder from academia/research asked how it is possible to have antibody response and reactions despite intestinal degradation of Cry1Ab. GRACE team members clarified that small amounts can still reach GALT; lots of proteins are, in fact, not fully degraded so it is possible to see potential response and published research show that Cry1Ab is quite resistant in physiological conditions. 67. Stakeholders from industry asked about validation and if metabolomics changes in feeding animals can be defined. It was also considered that the method is very sensitive and can be affected by many factors. A GRACE team member answered that values showing changes will be put together with all the other data for interpretation. Guidelines on compositional studies will be used. 68. A representative of CA asked on which material tests will be performed. GRACE team members replied that for the in vitro assays it is used the same feed material that is introduced for the animal diets of study A and B.

Session 4: Prospects and Networking Ralf Wilhelm, (Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany) presented the Further Studies (ppt slides see Appendix 3). He presented the chronic toxicity studies with MON810 maize (1-year, ongoing) design (compositional analyses, weighting, physical examinations and functional assessment, ophthalmology, haematology, biochemistry, urinalysis, necropsy and several preliminary results (compositional analysis, i.e. fumonisin found in control maize) and informed on the situation of the subchronic toxicity study with transgenic potatoes.

Christian Kohl (Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany) presented CADIMA – a new dissemination portal for information and data on ongoing and future biosafety research (ppt slides see Appendix 3). He illustrated the passive and active roles of the portal, the structure with 3 major areas, the GRACE contributions to CADIMA and the synergies with other projects on the areas of knowledge gap analyses, review of GM impacts, GMO info and toxicological studies.

Pablo Steinberg (University of Veterinary Medicine, Hannover, Germany) introduced the recently started FP7 project G-TwYST – GMO Two Year Safety Testing (ppt slides see Appendix 3). He provided basic information regarding the call, participants and the objectives of the project. He showed the planned feeding trials: a) 90-day feeding studies with GM maize NK603, b) the combined chronic toxicity/carcinogenicity feeding trials with GM maize NK603 and c) the carcinogenicity feeding trial with GM maize MON810.

Discussion 1-year chronic feeding study 69. A member of the academia/research community asked if a plan of the 1-year feeding study is already available and which is the rationale for conducting the 1-year feeding study since results of the 90-day feeding studies are not available yet.

19

A GRACE team member explained that OECD and EFSA normally ask for a trigger for this type of study, i.e. the results of the 90-day study. Corresponding with a mandatory 90-day study, no trigger would be needed. In order to fit into the timeline of GRACE, these studies are, however, not performed in sequential order. Results from the 90-day and 1-year studies will be compared to see e.g. whether the pattern of parameters observed is similar and/or whether the 1-year trial is providing any more information or not. 70. Stakeholders from industry asked which conclusions can be drawn on the two studies if only one variety of GM maize is used in the 1-year feeding study compared to using two varieties of GM maize in the 90-day studies. They added that differences between the 90-day trials A and B were already presented and the 1-year trial is partially a repetition of the 90-day trials with material that is contaminated. It was highlighted that industry would never start a study with this material because the statistical power changes and they wonder whether this study can really clarify the results of the 90-day studies. A GRACE team member answered that data of corresponding varieties can still be compared. It will be possible to see whether the patterns of the 90-day feeding trials are confirmed in the 1-year trial and whether the 1-year feeding trial is more sensitive and add resolution compared to the 90-day study, e.g. it might be investigated if the generic pattern seen in the 90-day studies becomes clearer. Fumonisins are present in the material. Therefore, it will be analyzed whether the observed changes might be related to the presence of fumonisins in the diets. 71. A representative of academia/research community asked why wasn’t used the same feed as for the 90-day studies and why mycotoxin levels were not determined in advance of the study. This would avoid raising more questions than answers at the end of the project. The GRACE coordinator answered that initially 90-day feeding trials were planned and the material was planned to be grown in Spain. To keep up with the timeline of the project, it was necessary to start growing the maize before the EC signature of the project. For the 90-day feeding studies only few kilograms of maize were grown in Spain (8 varieties) and dried with regular small machines; for the extended feeding studies, for which at least 4 tons of material were necessary, more material needed to be grown and dried in bigger facilities than the ones used for the 90-day feeding studies. The team therefore faced new challenges (limitation of the facility in Bratislava and financial constrains). Further analysis were conducted on the maize and it was decided to use MON810 grown in Spain; the idea was to use the same for G-TwYST and GMO90+; however, contamination were found in a batch (fumonisine) and it was impossible to use the same maize for G-TwYST so it was decided to grew new material for the G-TwYST and GMO90+.

CADIMA 72. A CA representatives required clarification on the type of data available via CADIMA, on when they will be accessible and what will happen to the data after GRACE comes to an end. A GRACE team member clarified that CADIMA is already accessible and that it will be filled as soon as data will be available. Raw data from feeding studies will be made available once the journal manuscript has been accepted for publication. Maintenance of the portal after the end of GRACE will be conducted by JKI with public funding. 73. A question was asked if meta-analysis is foreseen. GRACE team members replied that for meta-analysis quantitative data are needed, while other methods exist, such as mapping. Besides the particular reviews carried out in the project, CADIMA will also provide tools for reviewing for the use by any other party.

20

74. It was asked if omics data would also be in the CADIMA database. GRACE team members responded that the data will be included. Comment by GRACE team members: Industries have been contributing very much with their participation to GRACE events and the comments sent in different stages of the project; however, another help could come from their side. GRACE is also conducting Evidence Synthesis tasks and in the applications for placing on the market GM plants many data are included. Legally, the GRACE team have access to these data after presenting a request to EFSA, but the team is not allowed to use these data for the review process; it would be very helpful if EuropaBio could help the GRACE team performing SRs by making available these data. 75. Industry responded that it will consider the request.

G-TwYST 76. A member of academia/research community asked if EFSA addressed in its recommendations on chronic or subchronic toxicity and carcinogenicity and if the recently started French project [GMO90+] has been taken into account. A GRACE team member answered that EFSA has published a statement on the applicability of the OECD Test Guideline for chronic studies with chemicals to trials with whole feed. He also explained that there is a strong collaboration between G-TwYST and GMO90+. 77. A representative from EFSA clarified that EFSA published in July 2013, upon request of the EC, a scientific report on the applicability of the OECD Test Guideline 453 to combined chronic toxicity/carcinogenicity studies with GMOs. 78. A researcher commented that G-TwYST is a consequence of a study2 and it is unnecessary to perform it with EC and public money. He wonders if the consortium expects to find any toxicity and carcinogenicity and which is the rationale to continue the 1-year study in GRACE now that this is also done in G-TwYST. A GRACE team member explained that the consortium does not know if toxicity will be observed; therefore it is normally referred to as potential toxicity. These studies will be performed upon request of the EC with an adequate study design to avoid the mistakes of previous studies. 79. A representative of CA expressed concern on the 2-year study with MON810, specifically on the limited number of groups (absence of conventional comparators), which would affect what one can interpret. It was suggested to insert a historical control or to drop the study. A GRACE team member replied that the consortium has the same concerns, but it was not possible to include another experimental group because of animal housing and financial constraints. 80. A stakeholder from Industry asked which MON810 lines will be used. A GRACE team member answered that a commercially available variety of MON810 will be used for the 2-year study. 81. A stakeholder from academia/research asked if the quality of the test material will be checked, specifically the concentration of Roundup will be determined. A GRACE team member answered that it will indeed be checked.

2 Seralini et al., 2012. RETRACTED: Long term toxicity of a Roundup herbicide and a Roundup-tolerant genetically modified maize, Food and Chemical Toxicology, Volume 50, Issue 11, Pages 4221-4231.

21

82. A representative from the EC asked if it is possible to collect samples that might be used in the future for further experiments. A GRACE team member answered that it is already planned to put aside as much material as possible for possible studies in the future. Material will e.g. be shared with the French GMO90+ consortium. 83. A representative from CSOs/NGOs asked where the NK603 maize come from. A GRACE team member answered that the source will not be disclosed, but that it will be legally obtained.

Collaboration with EFSA 84. A participant asked about the collaboration with EFSA. A GRACE team member explained that close collaboration with EFSA was required by the call text. A GRACE team member added that GRACE has to provide reliable data for the review of the EU Regulation on applications for authorization of GM food/feed. Revision of that Regulation will be based also on the results from GRACE, which is doing the best to provide good and transparent data which had been discussed with stakeholders. 85. An EFSA representative explained that EFSA is preparing an explanatory note on 90-day studies. An important point would be a clarification and to strive towards greater consistency among studies provided. The EFSA guidance allows for flexibility. Interpretation and implementation of guidance allows for some flexibility. This is still a theoretical exercise and intended to cover the period until the end of the GRACE project, providing practical indications, particularly when there are no hypotheses.

Other questions and comments 86. An industry representative sought clarification on issues like number of animals, randomization, combination of sex groups; what is the trigger for the feeding studies and which are the endpoints; if knowing which toxicity pathways are activated could be the trigger for a 90-day feeding study. A GRACE team member explained that the results from the in vitro studies might be the trigger for the feeding studies. By now a more complete answer cannot be provided. Knowing if a pathway is already activated is a good starting point to decide if to perform feeding studies. So far, GRACE is working on a step-by-step approach. In vitro studies might be a starting point: if something a toxic effect is revealed observed in these studies, then feeding studies could follow, if deemed of added value, starting with 90 days and then 1 year or 2 year trials, depending on the findings. One of the open questions is what effect sizes should be established. 87. A member of the academia/research community commented that the consortium is in a unique position and it would be very good if recommendations about design in relation to statistics at the end could be made. It is important that methods are standardized and validated. 88. An industry representative commented that the feeding studies lack specific objectives and his concern is that this project will lead to greater regulatory burden not based on scientific evidence. It would be better to use the budget of GRACE for e.g. in vitro methods. 89. An EC representative responded that very good results were already reached in GRACE. The outcome will be beneficial not only for consumers but particularly for the industry. Industry

22

should consider that not all issues can be addressed by GRACE. Industry might contribute additional money for further studies. 90. A member of the academia/research community asked about the consumer benefits and the objectives: what is the rationale to look for unintended effects of GM crops specifically as compared to conventional breeding. As example, there are new rice varieties with considerable changes and a likelihood of unintended effects. The GRACE coordinator explained that assessing conventional crops is not within the scope of the project. While this is a scientifically valid remark GRACE has to be seen in the context of the EU GMO legislation. According to the EASAC report, the current risk assessment and regulation are not evidencebased; the focus should be on new trait and product independent from the technology used to introduce them. 91. Industry representatives responded that GRACE is a publicly financed project with no influence from industry. Hence, it is contradictory asking for a financial contribution from industry. Moreover, industry always provide comments through EuropaBio but the general feeling is that they are not always addressed and answered. It was added that industry needs a very good hypothesis to finance a study and due to the focus on regulatory issues of this project, industry might be not the best category to ask for funding. The GRACE coordinator answered that many stakeholder groups are involved in the dialogue besides industry and industry’s comments are always much appreciated. They are seriously considered and everything is documented. A GRACE team member added that all the issues will be addressed and, if some documentation or response from the GRACE team are not available yet, it is only a matter of time: documenting and processing all questions and comments received is very time and resource intensive. The flexibility of GRACE to consider comments and suggestions is also constrained by the rules and budgetary limitations of a FP7 research project. A GRACE team member commented that contribution from industry does not necessarily mean money. There are lots of data in applications that GRACE is not allowed to use but it would be very useful to have access to them. A GRACE team member specified that, of course, GRACE has access through EFSA to look at the data but they cannot be used. It would be good if EuropaBio could help GRACE in using these data also in the course of the GRACE systematic reviews. 92. A CSO representative asked for more data on the in silico results. It was asked to state clearly at the end of GRACE if feeding studies are really needed. A GRACE team member explained that this item is covered by Work Package 2. In vivo and in vitro will be compiled to build an in silico prediction tool at the very end of the project. 93. A representative of the EC concluded that this was a very useful workshop. The EC is looking forward to the final results of GRACE which will be discussed with Members States during the revision of Commission Implementing Regulation (EU) No 503/2013 on applications for authorisation of genetically modified food and feed. 94. Members from academia/research community commented that identifying triggers is more important than hypotheses. Hypothesis is needed for defining the scientific rationale of a study design. It was added that reduction of uncertainty should be the aim of the studies. GRACE team members responded that there should be a scientific rationale for any feeding study.

23

Next Steps and closing remarks The GRACE coordinator concluded the workshop highlighting the possibility to provide further comments in writing in the following two weeks. The manuscript describing the results of the two 90-day feeding studies will be submitted in 4 weeks. The idea is to publish it as open-access journal paper and GRACE is negotiating with the publisher to include also raw data as supplementary data. If this will not be possible, the data will be published on the GRACE website and available to stakeholders. The journal will also provide for a discussion forum. Any comment, criticism, discussion will be appreciated. In the second half of 2015, a third stakeholder workshop on the results of feeding studies and another stakeholder workshop on the results of the systematic reviews and evidence maps will be held. He thanked the EC, the WP7 staff, all GRACE partners and all participants for contributing to the meeting.

Written Comments and Responses of the GRACE Team A total of 10 organisations or individuals resp. provided written comments after the stakeholder workshop. These comments were categorised according to topics and discussed by GRACE team members in order to what extent these comments can and will be adopted when finalising the interpretation of the results (for the animal studies) in proceeding with work (for the alternative studies). Responses to each of the comments were developed. Both stakeholder comments and GRACE team responses are provided in Table 1. For the fully referenced original stakeholder comments – as they were received in writing - see Appendix 4.

24

Table 1: Stakeholder comments received in writing and responses of the GRACE Team*

No

Stakeholder Comment/Question

GRACE Team Response

Considered in publication*

Considered in a later phase*

Cannot be considered*

General 1

Given its assignment and composition, the GRACE project (no Q; statement) consortium is in a unique position to help remind/clarify a number of important characteristics of Environmental risk assessment (ERA) and food/feed safety assessment (FSA) of GM crops in the context of regulatory decision making: 1. The principles and methodology of ERA and FSA are based on internationally developed approaches such as Annex III of the Cartagena Protocol on Biosafety (CPB) and guidelines of the Codex Alimentarius. 2. While ERA and FSA have different assessment endpoints and characteristics, they share that they need to be conducted in a manner that is: a. scientifically sound and transparent b. well structured, stepwise c. tiered and hypothesis driven, i.e. early stages in the assessment of a specific assessment endpoint (e.g. toxicity) indicate whether and what additional information or testing may be required, i.e. those next steps have a scientific rationale ( ‘hypothesis driven’ ). d. Additional data should only be asked when one knows what to do with the answer

*

In case of questions where there is no information included in all three cells containing information how the comments have been processed (three columns on the right), this indicates either that the comment is of a general nature or that this information is missing from the GRACE team. What is meant by publication is scientific paper reporting on the results of the 90-day feeding trails which was already available as open access at the time of the publication of this report: Zeljenková D1, Ambrušová K, Bartušová M, Kebis A, Kovrižnych J, Krivošíková Z, Kuricová M, Líšková A, Rollerová E, Spustová V, Szabová E, Tulinská J, Wimmerová S, Levkut M, Révajová V, Sevčíková Z, Schmidt K, Schmidtke J, La Paz JL, Corujo M, Pla M, Kleter GA, Kok EJ, Sharbati J, Hanisch C, Einspanier R, Adel-Patient K, Wal JM, Spök A, Pöting A, Kohl C, Wilhelm R, Schiemann J, Steinberg P. (2014). Ninety-day oral toxicity studies on two genetically modified maize MON810 varieties in Wistar Han RCC rats (EU 7th Framework Programme project GRACE). Archives in Toxicology 88(12): 2289314. NA: not applicable; Q: question 25

No

2

3

4

5


e. comparative - identified potential adverse effects are considered in the context of the risks or effects posed by the nonmodified recipients and their use. Policy advice: As indicated in the GRACE consultation report (March 2013) this project is expected to provide policy advice: ‘The aim is to evaluate the EFSA guidelines in order to provide a scientific basis for policy development and to define gaps in RA’. As communicated in previous consultations, EuropaBio is concerned about the fact that the GRACE project seems to be adhoc tailored to respond to policy-making demands and would like to emphasize that the aim of the Research Framework Programme funding should be to support research and innovation with the final end of reaching the goals of growth, competitiveness and employment. Study rationale: As raised during previous consultations by EuropaBio and other participants, the rationale for conducting the studies and the main objective of the project was not considered in sufficient detail. We would like to emphasize that the outcome of the study should include recommendations on how to conduct 90-day feeding studies. Following more than 150 studies in animals with food produced from GM plants, which were substantially equivalent to their non-GM comparator, no unintended biologically relevant effects related to their safety as food were detected (Flachowsky 2012; Snell et al., 2012). Use of data for the CADIMA database project: EuropaBio will consider and discuss internally if animal feeding data submitted in applications can be used for the CADIMA database project by GRACE for systematic review process. Submission of 90-day feeding studies in previous applications: EuropaBio would like to clarify that 90-day feeding studies have been submitted by industry in the past, without a trigger but for expediency during the regulatory process as some Member States were asking for it in any case. However, evaluation of the voting

GRACE Team Response




(no Q; statement)

The purpose of the feeding studies is indeed to consider an appropriate design and moreover the rationale for conducting subchronic and extended (chronic) feeding studies with regard to the risk assessment of GM crops. The studies are not undertaken to question risk assessment or consent for applications.

GRACE appreciates the support by EuropaBio. A communication procedure with the data owners has been established via EFSA and EuropaBio. (no Q; statement)

26

No

6

7

8


behavior by Member States reveals that they vote the same way on product approvals regardless whether 90-day rat studies have been provided or not. EuropaBio members are of the opinion that when substantial equivalence (i.e., no biologically relevant difference) is shown between the GM crop and its conventional counterpart, there is negligible value in the performance of animal feeding studies with rodents to assess food safety. This position is supported by international risk assessment experts who have concluded that whole food feeding studies are unnecessary on GM crops that have already been demonstrated to be substantially equivalent to their conventional comparators by the molecular, compositional, phenotypic, and agronomic analyses (EFSA, 2011 ; Kuiper, 2012 ). From an animal welfare perspective as well as the 3Rs commitment to replace, refine, and reduce in accordance with the European Commission’s Directive 2010/63/EU , the use of animals in scientific research should be minimized. International quality standards: Although the animal feeding studies are being/have been conducted according to the principles of Good Laboratory Practice (GLP), the generation of the plant material by the GRACE project does not seem to follow these principles. We would like to highlight the fact that this is not consistent with industry practice. According to the EFSA guidance and EU Commission regulations, toxicological studies have to fully comply with the principles of GLP or these will not be considered for scientific evaluation. Studies, other than toxicological studies, also have to comply with the principles of GLP or be conducted by organizations accredited with ISO standards. Given the purpose of the GRACE Project, and if policy decisions will be taken based on results from studies conducted under the GRACE project, the same level of scrutiny should be followed for the studies carried out by the GRACE Project as Industry would receive for one of its studies.

GRACE Team Response




Agreed. Therefore, a thorough consideration of study design and necessity as well as a consideration of alternatives is conducted within GRACE. Data generated throughout the GRACE project are not considered part of a regulatory risk assessment. The studies shall be critically reflected - e.g. within the stakeholder consultations - to critically consider study designs. Based on the studies and the respective discussions conclusions and recommendations will be discussed with stakeholders and finally published.

see 7; essentially the studies shall support critical scientific discussion on study designs and on the rationale for conducting the studies.

27

No


GRACE Team Response

9

Historical Background Data: According to the 2011 EFSA Food/Feed guidance and Implementing Regulation 503/2013, the use of historical background data to determine the natural variation of test parameters is highly advisable.

10

In the case of the 90-day studies, the chosen testing facility did not have any relevant historical data, so the GRACE team included two conventional controls that create a limited historical data set. In addition, we would like to highlight the fact that, in the case of the long term studies discussed during the stakeholder consultation, no conventional controls will be included (except in the case of the 1y study where one conventional control is included). A more robust historical background data set for the endpoints under investigation on the 90-day study is available from the animal supplier . We encourage GRACE to use this third party data to assist in interpreting the biological relevance of differences observed between treatment groups. Studies with laboratory animals are difficult tasks. I am sure that as leaders of the GRACE projects you have in mind that the animal model (In House Historical Data) and the expertise in conducting those studies with appropriate equipment and software) are crucial. There are some questions about the experts involved in this part of the GRACE project that need clarification: Apparently, Pablo Steinberg from the University of Veterinary Medicine Hannover is now a core member of the GRACE project team. However, there is no mention of Hannover University as a GRACE partner. Further, the Bundesamt für Verbraucherschutz und Lebensmittelsicherheit (Federal Office for Consumer Protection and Food Safety) is now a GRACE partner. The role of this agency in GRACE remains unclear.

OECD guidlines suggest to use historical data only when generated in the same lab and within the last two years. Considering this rule GRACE tries to generate "historical background data" by additional maize varieties as far as it is possible within the project. see 9)

11

12

13




see 9); The database provided by HARLAN will be taken into account.

agreed; see 9)

Pablo Steinberg joined the "Task Force" of GRACE - a board of external experts. It should ensure that the project gains access to renowned expertise in toxicology. BVL is partner in work package 5 and conducts reviews on environmental impacts of GM crops.

28

No


GRACE Team Response

14

How will conflicts of interest be addressed and communicated in the case of Kerstin Schmidt, whose company, BioMath has extensively collaborated with Monsanto? For example, BioMath has for many years been involved in the preparation of Monsanto's post market monitoring reports on the cultivation of maize MON810 in the EU (see for example: http://ec.europa.eu/food/plant/gmo/reports_studies/docs/repor t_2012_mon_810/report_2012_mon_ 810_farmer_questionnaire_survey_en.pdf). The financial support of this project by the European Commission is highly appreciated and warmly welcomed by Hungary. The idea of providing financial resources for studies not financed by the notifiers themselves in order to gain independent experimental data on the safety of GM crops is of utmost importance for us and for many other Member States. Projects like this might also relieve the pressure of consumers for proper long term experiments and strengthen scientific risk assessment. Great care was taken by the experts of the Toxicological Institute of The University Group of Bratislava to carry out the in vivo animal experiments. The work of the Slovak colleagues, who "blindly" worked for several months to provide data without any bias, is fully appreciated. Their work and diligence is commendable to all. We advise to the GRACE Management Team, the EU Commission and EFSA that the 90 day feeding studies should be repeated by control diets free from any GMO contamination. We are convinced that the performance and outcome of these studies will be given much attendance on behalf of the EU and international scientific, industrial and consumer fora. Also for that reason, special care should be taken when performing these independent studies in a most transparent way.

The interests of BioMath, respectively Kerstin Schmidt, will be declared accordingly. BioMath is responsible for the statistical analyses of the data generated. It is not involved in the practical perfomance of the studies. Any third party will be able to analyse the data: The study data are published open access, made accessible via CADIMA (www.cadima.jki.bund.de), or large data sets > 1 GB and slides can be accessed on request from the respective project partner. (no Q; statement)

15

16




(no Q; statement)

29

No


GRACE Team Response




90-day feeding studies 17

18

19

Design and Analysis The GRACE consortium re-iterated that it is too early to draw any We agree with the comment. conclusions from the results that would impact risk assessment for GM crops. We would like to emphasize that the main objective stated by the GRACE project is to provide further guidance on how to conduct a 90-day feeding study with GM material and not to open the discussion on the safety of GM (MON810).

Interpretation of existing 90d studies: The fact that, for certain endpoints, the order of assessment was systematic, with replicates of each entry being assessed in groups of four rather than being fully randomized, has the potential to introduce bias which could lead to apparent differences between genotypes that are in fact caused by temporal or other, non-treatment-related trends. This may explain why certain differences were apparently detected in Study 1 but not Study 2, and vice versa. Design and conduct of future 90d studies: The two 90d studies both had sample sizes of 16 animals per group, which was chosen on the basis of achieving a certain probability of detecting a standardised effect size equivalent to one standard deviation. A key drawback of this approach is that standardized effect sizes take no account of toxicological relevance. An extensive review carried out by EuropaBio members in 2012 led to the conclusion that a sample size of 10-12 animals per group was sufficient to provide high power to detect toxicologically relevant effects across a range of key endpoints. A sample size of 10-12 animals per group therefore represents a more appropriate balance

This aspect will be taken into account when working out the conclusio ns of the GRACE project.

The measurements of the different endpoints in the different groups occurred within short to very short periods of time, so that the factor time most probably had no influence on the outcome of the studies.

This aspect was not taken into account in the study plan.

The sample size calculation was based on the standardised effect size (SES: the difference in means between control and treated groups divided by the standard deviation [SD]). It was assumed that an SES of 1.0 SD or less is unlikely to be of toxicological importance. Assuming a power of 0.8, a significance level of 0.05 and a two-sided test this would be achieved by 17 animals per group. The number was rounded down to an even number (i.e. 16) as the rats were to be housed in pairs. The number of animals used in the 90-day feeding trials does not contradict the comment by EuropaBio.

This aspect was not discussed in the paper by Zeljenkova et al. (2014).

30

No

20

21


between power and ethical considerations regarding the use of animals in research. It is important to highlight that study plan parameters for 90-day feeding studies in EU studies might impact industry’s studies submitted in many other countries outside EU and set new standards. An example of this is the inclusion of pancreas weights on the GRACE 90-day studies. Pancreas weights are not required by OECD TG 408, and are not among the organ weights considered to contribute the most value to safety testing conducted with foods (Michael et al., 2007). Thus, the reason for including them in the current studies has not been sufficiently explained, and in the absence of a scientific rationale it should not create a new regulatory requirement.

Hungarian experts and the Hungarian Competent Authority are still not in full agreement with the design and construction of many of the scientific studies of the Grace Project. The approach applied differs on several points, sometimes significantly. Our main concern is that the two 90 days rat feeding experiments have not been carried out properly. The main problem is that the

GRACE Team Response

We agree with this comment: the decision to include the pancreas weight was a sole decision of the colleagues at the Slovak Medical University. It does not match with internationally accepted guidelines and was not included to create a new regulatory requirement. Since the pancreas weight was determined, the consortium agreed that in terms of absolute transparency the data had to be communicated.

We totally disagree with this comment: the minimal contamination of the diets with GMO had no influence on the outcome of the 90-day feeding trials. Moreover, the fact that the studies were performed in a blinded manner made sure that even in the presence of a minimal GMO




The pancreas weights were included in the paper by Zeljenkova et al. (2014), and this issue was discussed in a response (Archives of Toxicology 89: 137139, 2015) to a report and press release by BauerPanskus and Then (2014). X

31

No

22

23


diet prepared from MON810 GM maize and its isogenic control line, as well as the commercial control maize varieties were contaminated with GMOs. Since these contaminated control maize diets were colour coded and used for the two 90 day toxicological/feeding experiments, and the personnel carrying out the work was unaware of this contamination, the results gained with these two experiments are doubtful. Animal diet and dosing Maize cultivation: The cultivation of the GMP and the nearisogenic line took place in a certain distance to each other. The reason for this should be given and the possible influence on the composition of the plants should be discussed. It could lead to a higher variability between the lines because of different conditions and therefore cover effects caused by the genetic modification.

GRACE Team Response




contamination of the diets this contamination did not influence the performance and outcome of the study.

Thanks. The comment points to an issue also previously discussed, namely that on one hand, the cultivation of maize MON810 in relatively large plots (still within each other's vicinities) and quantities served the purpose of producing sufficient, homogenous material with the least likelihood of cross-contamination with e.g. pollen-flow from neighbouring fields or preventing admixture during harvesting, while usually field trials for compositional analysis follow a different design with relatively small plots and a block design in multiple blocks in multiple locations (allowing also for extensive statistical analysis). The prime objective was to produce high-quality materials for feed that had to be well-characterized; the different plots were within each other's proximity and agronomic practices applied to them were the same. We acknowledge that this of course won't allow for a full-fledged comparative analysis as recommended for dossier studies, yet are of the opinion that the compositional analysis provides useful insights into the similarity of the feed materials. Diet: As it is not expected that GMO will be highly toxic, the GMO- This issue has indeed been discussed, for example with the content should be as high as possible. A content of 33% is not the nutritionist of the company producing the tailor-made highest possible concentration without putting an extra stress on rodent feedstuffs. It was indicated that for maize to be the rats and is therefore not optimal to find effects. Also the included without impacting on rodent nutrition, one could presence of GMP in control maize varieties can mask effects go perhaps up to 40%, which is not dramatically increased as compared to 33%. Moreover, as 33% maize inclusion 32

No

24

25


GRACE Team Response




caused by the genetic modification and therefore the relevance of appeared to be standard practice for standard rodent diets the outcome of the study can be questionable. produced by some producers and is also frequently used in published literature on GMO feeding studies, this would also allow for comparison of study outcomes with other historical data and previously published studies. As only minute concentrations of the active principle (i.e. the The aim of whole-food testing (i.e. for unepxected transgenic protein) are present in the diet a very robust and unintended effects of genetic modification) is different precise study design with a minimum of biological deviations and from testing the safety of the newly expressed protein bias is mandatory to discover differences in the effects mediated (which can be done through toxicity tests in which the by the GM maize containing feed compared to the study group animals receive the protein in purified form). Various of exposed only to the near-isogenic non-GM comparator containing the issues discussed with stakeholders, such as statistical feed. power versus number of animals, standardized effect size and other considerations (design, rat strain, feed, etcetera) take into consideration the exclusion of confounding factors and ensuring high sensitivity of the animal study performed. GM maize concentration of 2,6% in the conventional control 1 of Thanks for pointing at the issue. There was indeed study B is inacceptable as non-GM comparator. background presence of MON810 event in maize from the control maize (genetically close to the GM) and the two commercial varieties used in each study. Such presence is regarded as adventitious, which can relate to crosspollination in the field (maize MON810 is widely grown in Catalonia), traces present in seedlots or commingling at other, postharvest stages, while measures had been taken to avoid such commingling to the extent possible (e.g. equipment cleaning and order of processing during feed production). The background presence that the commenter refers to is in two of the four commercial lines that had been used to create a background of values of animals fed diets based on commercial maize ingredients. A level of 3% MON810 means that 97% is non-MON810 and therefore if MON810 were to have an impact as compared to commercial maize, the comparison of 33

No


26

The composition of diets and the growth patterns of individual animals will naturally vary some between crop lots and animals. You have a very restricted number of samples. There is some variation, for instance in total protein and even more so in fructose and glucose and total carbohydrates for some of the samples (Pioneer GM and near isoline). One might expect there to be some difference in pancreas weights. There was also a high bacterial count in the GM and near isoline of the A study grain and feed. Wouldn’t you expect that to have some impact on some immune measures? Was the LPS content measured?

27

The data on feed consumption during the last 2 weeks (12 and 13), and especially the last week (13) show an unexpected drop in consumption essentially across the board. What is the case? Is it that the final week was a short-week? Maybe 6 days instead of 7? Was it because the feed was becoming rancid? Was it because some key component, like a vitamin was declining? Or was it due to a shortage of water? Clearly it also did have a minor impact on body weight gain as well. Were there measurements of the feed at the end of the study to determine if there was spoilage (yes, I realize the diets were irradiated), it seems that rancidity or nutrient decomposition is a reasonable thing to question. Otherwise, water supply.

GRACE Team Response




MON810 (100%) with these two groups would still be expected to show this impact. The commenter correctly notes that various parameters showed different values in the analysis of maize. When checking the analyses of diets derived from these maize varieties, the contents of proteins and various other parameters considered were rather similar and therefore not highlighted further in the discussion of these parameters in the summary report of the feed analyses. The source of the enterobacteriaceae counts might have related to their presence in irrigation water as no organic manure had been used for the maize grown for this study (counts reduced to zero after irradiation). As regards lipopolysaccharides, these are known to be of concern following systemic exposure, such as through interperitoneal or intravenous injection in laboratory animals or through inhalation of dust in workers, yet it seems that oral exposure does not have adverse effects on animals (e.g. http://ar.iiarjournals.org/content/31/7/2431.full.pdf ) No observations of feed deterioriation had been noted by the staff feeding the animals, and feed had been stored in the facility under controlled conditions not expected to lead to such deterioriation (these pelleted feeds are relatively stable and they have been irradiated). No analyses of feed had been undertaken at the end of the study, yet samples were retained to be able to do so in case this is needed

34

No


28

Feed preparation: Was feed analysis performed under GLP?

29

Treatment of maize and Bt content: The draft for publication says the maize was milled in Italy, but in his presentation Mr. Kleter says it was milled in Wageningen and there is some mention of pellets in this context. Clarification is needed regarding this process.

30

What was fed to the animals? Pellets or milled maize? Which temperature was used at which level of processing? Please also describe the process of radiation that was used.

31

Which protocol was used in the analysis of Bt content? How was this method evaluated to make it comparable to other publications/ data? At what stage of the experiment did you determine the Bt content?

GRACE Team Response




The laboratories involved included Covance (implementing GLP and quality assurance / quality control procedures for the analyses performed for GRACE) as well as Mucedola and RIKILT (working in compliance with ISO 17025) and INRA (for the analyses of Cry1Ab protein levels, working according to principles of GLP although not particularly accredited for this method) The maize was indeed milled at the animal feed producing facility in Italy (before preparation of the diets made from it) while for analysis, diet samples (consisting of pellets) had to be milled (in Wageningen, after samples of the diets and the maize used for feed preparation had been shipped there) Pellets were fed to the animals. A facility for tailor-made feeds was used at the ammal feed producing facilitiy in which elevated temperatures were avoided to the extent possible (e.g. using a pasta press after mixing basal mix with other ingredients, cooling by adding water in the process) and drying of pellets thus produced at 50°C for a limited time (until moisture has been removed), followed by vacuumpacking. Irradiation was done in a certified gama irradation facility (approved according to EU Directive 1999/2/EC) where products are put on carts of an automatic system moving the carts around the irradiation source (see for more details, including text, video and map at http://www.gammarad.it/it/processo.php ). The diet batches thus produced were succesfully irradiated at a target dose of 1.5 MRad (standard dose). Two methods have been used to confirm the presence of MON810 maize: PCR according to methods ahering to e.g. mnimal performance requirements of the European Network of GMO Laboratories, while the Cry1Ab method was done according to the method published by Adel35

No

32

33

34


GRACE Team Response




Patient et al. (2011; http://journals.plos.org/plosone/article?id=10.1371/journ al.pone.0016346 ) Comparison of plant composition: There are a number of This issue was also raised previously during the differences in the composition of the two MON810 varieties and preparatory workshop in Vienna, December 2012. the isogenic varieties (trypsin inhibitor, protein, carotenoids...). Because the field production had to be dedicated to These differences should be translated to the EFSA equivalency production of relatively large quantities of maize for categories to find out whether one could assume equivalence. animal feed production, it was decided not to use a field trial design as recommended for the comparative compositional analysis of GM crops (e.g. as EFSA does) given that this would require a different more complicated layout with lesser quantities produced per plot and enhanced risk of cross-contamination before, during and after harvest. The composition of both the maize itself and of the diet prepared This is an issue also disscussed at other occasions. During from it would allow for 55-66% inclusion of maize in all the diets. the preparations of the feeding trials, the project The higher the dietary inclusion, the more effect can be detected. consortium had an intensive dialogue with the nutritionists In spite of this, the example of industry using 11 and 33% was of the animal feed-producing company. Their feedback followed. The Hungarian Competent Authority disagrees with this was that up to around 40% of maize could be incorporated approach. into the diets without any major nutritional impact. This elevation was considered relatively small. Moreover, using the 33% level would allow for comparability with other experiments (NB the reason why 33% had been chosen by other authors was that this was apparently a commonly used inclusion rate of a particular large laboratory animal feed-producing company for their standard diets, which, in turn, would allow for comparison of historic data of rats fed standard diets). Relevance of traces of other GM events In the analysis of the test material, it is referred that slight MON810 was detected in two of the four reference diets admixture of GM maize was noted in some maize used in the (made of commercial maize) used for additional study; can you please comment on this, in particular clarify if the comparisons besides the primary comparison of groups fed source of contamination has been identified, what are the MON810-containing diets with their direct controls. In 36

No


GRACE Team Response




implications for this study, if any, and which other considerations can be done on this issue in general?

35

these cases, the results of both DNA- and the Cry1Abbased assays indicated that quantifiable levels of MON810 were roughly 1 order of magnitude lower than the lowest tested level (11% w/w diet) of MON810, indicating that most (>>95%) of the maize in these diets was still nonMON810. One of the control diets contained traces of MON810 of which levels were too low to allow for their quantification (hence orders of magnitude lower than the tested levels of MON810), which relates to the high sensitivity of the detection method. Such low level is not considered to impact on the appropriateness for comparisons with diets containing 11 and 33% MON810. As the presence of MON810 in diets correlated with that in the maize materials used for feed preparation, it is obvious that the admixture must have happened before feed production. We have not been able to pinpoint the source (examples of possible sources: sowing seed containing GMO traces, cross-pollination), while certain measures (e.g. field trial design, order in which maize batches were handled, cleaning of production equipment) had been taken to avoid cross-contamination. There is a realistic chance of low-level presence of GM materials in commercial practice and hence this may not always be avoidable in spite of these efforts. We identified several other problems, which may undermine the The comment only refers to the presence of GM materials confidence in these very important EU-financed studies and its but does not indicate a particular reason why this should scientific standards. As mentioned earlier, the non GM control be a problem. As explained above in more detail, the low, maize varieties/diets (both the isogenic- and the reference sometimes even not quantifiable levels meant that most of varieties) were contaminated with GM events. The contamination the maize present in the diets was actually of non-GM was above the limit of detection, in one case it was near to 2% nature and therefore the comparisons are not implicated (commercial line I). In another case the contamination originated from an event which has not been authorised for cultivation in Europe. Although the Hungarian Competent Authority is fully 37

No

36

37

38


aware of the difficulties, we still believe that the experiments should be repeated with new feed materials, which are not contaminated with GMOs and are up to the best scientific standards. Statistics I have only a technical question for you, regarding the periodical observation on rats and its statistical analysis. In your excel file, in data used for statistics there is the gain of body weight for each cage per day. Why did you considered not the week but the day? Is there some reference for this? Or it is because for the long term study you have to evaluate the food efficiency ratio and you probably will do this ratio on day and not week? Why did you decide to not plot the gain? Where is the statistic report for feed consumption (you mentioned the analysis in material and method page 9)? By inclusion of additional conventional maize varieties into the feeding scheme naturally occurring variability between conventional maize varieties and their effect on the consumer can be determined. However, including too many collectives, clusters, controls, and in this specific case even verum varieties increases statistical variances and reduces the statistical power of the analysis. This approach may obscure significant effects induced by the genetically modified feed constituent. An example is the significant and even dose-dependent diminution of the pancreas weight with male rats in feeding trial A. This effect is completely eliminated if the data of both feeding trials A and B are merged and statistically evaluated in a combined fashion. We recommend putting the focus on results obtained from animals fed with the GM variety compared to the study group receiving the nearisogenic non-GM comparator containing feed. I believe I made this comment with or at the Berlin, Germany meeting of GRACE. I believe the study design should have included a number of known toxic or anti-nutrient components to

GRACE Team Response




Actually, for analysis we recalculated the weight gain per week and used this data for analysis. Weight gain is plotted in Statistical analysis report feeding trial A_B, Appendix 5, chapters A2 and B2 Statistic report for feed consumption: Statistical analysis report feeding trial A_B, Appendix 5, chapters A3 and B3

In risk assessment of GMO it is common agreement to assess a GM plant with regard to the biological variability of conventional varieties, all with a history of safe use (see EFSA guidance for for the analysis of compositional, agronomic and phenotypic data). The intention is rather to determine whether the GM plant measures are outside of the commonly accepted range of values than to calculate any single differences between varieties.

The study plan did not include such experimental groups due to financial and animal housing limitations. Moreover,

X

38

No

39

40

41

42


help establish biologically significant differences for many of the parameters. That is even more important in the new general nontargeted “metabonomic” studies that have just begun. Looking for “statistically significant differences” in a non-targeted way, or even for the 90 day or chronic rat feeding study blood chemistry and organ weights should identify a number of random differences, as the current data (mostly) seems to show. Therefore these studies lack the most important group(s), positive controls to demonstrate biologically important variation. Since different toxins and anti-nutrients have different modes of action, one control is not enough. If these studies are to be useful, there must be a biologically relevant basis for setting standards of acceptability. If as in these studies, you are making a large number of measurements from studies with feed that we are pretty darn sure do not have anything wrong in terms of safety or nutrition, what do you think you are measuring? Maybe you are only getting close to measuring the inherent variation in the maize, the diets and the rats. That is NOT a safety issue. The sample size used in the experiment, that is the number of experimental units used per sex and per group,was calculated on the basis of SES; in particular SES=1 or less was judged to be of no/little toxicological significance by the Consortium. The number of experimental units in the study was then indicated to be 17 (rounded to 16) assuming a power of 0.8, a significance level of 0.05 and a two-tailed test. However, considering that rats were housed two per cage the number of experimental units should correspond to 8. Clarification is required on definition of experimental unit and on the actual number of experimental units/sex/group in both substudies A and B. More in general, it is considered useful to provide feedback on the appropriateness of SES=1 on the basis of the experience and data from this study (a posteriori analysis) and to possibly

GRACE Team Response




this issue was controversially discussed by stakeholders, some of them being totally against it. The study plan did not include such experimental groups due to financial and animal housing limitations. Moreover, this issue was controversially discussed by stakeholders, some of them being totally against it.

X

The careful analysis of the results performed by the GRACE consortium showed that there are no biologically relevant differences between the groups. In this case, the conclusion is that in the case of the genetically modified maize MON810 there is no safety issue.

X

The sample size n is calculated as a function of four parameters: the difference of interest (biological relevance), the variance or standard deviation in the population and the errors of first and second kind. In case values are averaged (animals are housed in cages) the standard error of the mean instead of the standard deviation is applicable. The number of singly housed animals consequently is the number of animals in housed in two cages times the square root of two.

This question is broadly addressed in the G-TwYST project.

39

No

43

44


GRACE Team Response

extrapolate further to support this approach in 90-day rodent studies inGM applications in the absence of a specific hypothesis. Statistical analysis of hematological and clinical chemistry data from OECD TG 408 studies on a “cage basis” instead of processing individual animal data is highly unusual, even though it may be in line with EFSA guidance on conducting rodent studies on whole food/feed. Taking into consideration the large amount of statistically significant differences of 11% and 33% GM groups, a re-analysis of numerical individual data should be performed.

Since the experimental unit is the cage all characteristics have been analysed on a cage basis, i.e. mean values of two animals instead of single animal values were analysed. Confidence intervals of averaged values are broader than those of single values, consequently when analysing individual animals more significant differences are expected.

Haematology, biochemistry and histopathology To detect any effects induced by genetically modified maize containing feed during this short period of exposure (i.e. 90 days) the tested endpoints have to be extended to include also immunological parameters like B- and T-cell subpopulations, the analysis of changes in membrane receptors of theses lymphocytes and the quantification of cytokines. Finamore et al. (2008) evaluated the gut and peripheral immune response to genetically modified (GM) maize in mice in vulnerable conditions.(1) Weaning and old mice were fed a diet containing MON810 or its nearisogenic control maize or a pellet diet containing a GM-free maize for 30 and 90 days. The immunophenotype of intestinal intraepithelial, spleen and blood lymphocytes of control maize fed mice was similar to that of pellet fed mice. As compared to control maize, MON810 maize induced alterations in the percentage of T and B cells and of CD4+, CD8+, γδT, and R T subpopulations of weaning and old mice fed for 30 or 90 days, respectively, at gut and peripheral sites. An increase of serum IL6, IL-13, IL-12p70, and MIP-1 after MON810 administration was

The feeding trials were designed by taking into account the OECD Test Guideline 408 and the corresponding EFSA statements, and in accordance with these no further immunotoxicity endpoints were determined.




Immunolo gical paramete rs will be determine d in the 90-day longitudin al study.

40

No

45

46

47


also observed. The results were indicative for the importance of the gut and peripheral immune response to GM crop ingestion as well as the age of the consumer in the GMO safety evaluation. In the present study no specific measurements/analyses were performed to validate the results of Finnamore et al. My main comment is that much more attention has to be given to the histopathology part of those studies with laboratory animals. I know that you are well aware that biological relevance of the findings is more important than the statistically significant changes. In such studies with laboratory animals, the interpretation of the toxicologic pathologist remains the central part of the safety assessment. The biological relevance of the findings is mainly given by the pathology report together with the clinical pathology conclusion. And the relevance of new tools such Omics still needs the phenotypic anchoring with histopathology data. Up to now, I consider that we got only a piece of information. During the last Workshop in Brussels, only a partial view of the data was given and the major raw data (histopathology data) were not available. I thank you for your proposal to have access to the slides. But it is not the role of the European Society of Toxicologic Pathology to review the slides. However we will be ready to advise you on that matter as we did when we emphasized the serious inadequacies regarding the pathology data presented in the paper by Séralini et al. (2012 ). Animal selection for haematology/necropsy workflow: it is not clear from the scheme presented if the sampling for haematology/necropsy was performed per group (i.e whole group 1, followed by whole group 2 etc.) or per cage (cage 1 per each group, followed by cage 2 per each group etc). Please clarify. Please provide information for clinical chemistry sampling workflow.

GRACE Team Response


We generally agree with the comment. Nevertheless, we would like to point out that the stakeholders have access to the complete set of data, not only to "a piece of information".

X

Half of the animals of each group was sampled on two consecutive days (see Table 2 in the Electronic Supplementary Material of the paper by Zeljenkova et al. [2014]).

X

Half of the animals of each group was sampled on two consecutive days (see Table 2 in the Electronic Supplementary Material of the paper by Zeljenkova et al. [2014]).

X



41

No


48

The incidence of histopathological finding presented is very low for a 90-day study in Wistar rat. It would be useful to have recent historical data illustrating background pathology from 90-day studies from the facility. It is recommended to involve the study pathologist and/or the peer reviewer in histopathology of the new 90-day study and in the 1-year study setting and evaluation. The dataset from the two sub-studies indicates inconsistent results in clinical chemistry and organ weights. This should be extensively addressed in the study report. Assuming that TOPALAB submitted a detailed pathology report, this report should be made available.

49

50

51

52

53

Quality of histopathological evaluation: OECD Test Guideline 408 requires the histopathological evaluation of all macroscopic changes seen during necropsy, independent of the dose group of affected individuals. This was obviously not done or at least it has not been reported. This omission becomes obvious when the gross necropsy findings (Table 10) are compared with histopathological findings (Table 11). For instance, inflammatory reaction is mentioned in gross pathology findings (animal Nos. 53 and 57, conventional group 1), but no histopathological correlate has been described. According to the presentation by Jana Tulinska, there were many significant differences in haematology and biochemistry which were partly sex specific. First, it should be clarified that sex differences must be considered very carefully as it has been shown that these are often underestimated in current research (see for example Clayton & Collins, 2014 ). Careful consideration should be given to the many significant differences in haematological and biochemical parameters. Most differences are not dose-dependent and only appear in one sex. However, it is quite obvious from the slides presented that there

GRACE Team Response


There is no historical data on Wistar rats in the facility. Regarding the second issue in the comment, a peer reviewer was involved in the 90-day studies and will be involved in the 1-year study.

X

This aspect has extensively been discussed in the paper by Zeljenkova et al. (2014).

X

The report in the Slovak language is available, and the individual slides can be viewed by any interested stakeholder. This comment does not apply: All macroscopic changes seen during necropsy were evaluated microscopically and the findings were included in the paper by Zeljenkova et al. (2014). The ultimate decision to determine if a finding such as "inflammatory reaction" in the necropsy indeed is an inflammatory reaction is taken based on the histopathological evaluation: in the mentioned case there was definitely no inflammatory reaction based on the histopathological analysis.

X

The differences were extensively discussed in the paper by Zeljenkova et al. (2014) and a comment by Steinberg (2015), and a sex-specific effect was not observed after a careful examination of the data.

X

There are significant effects in the GM groups, but as extensively discussed in the paper by Zeljenkova et al. (2014) and a comment by Steinberg (2015), these cannot be ascribed to the feeding of GM maize.

X



X

42

No

54

55

56

57


is an apparent accumulation of significant effects in the GM groups. In study A, significant differences were found in biochemical parameters in the 11 and 33% GM group each. No significant differences were found in the control. In haematology, significant differences were found in two parameters in the 33% GM group and in one parameter in the 11% GM group. No significant difference was found in the control. In study B, there are eight significant findings in biochemical parameters in the 33% GM group, six significant findings in the 11% GM group. There were no significant differences in the control. In haematology, there were significant differences in ten parameters and eight significant differences in the 11% GM group. No significant differences were found in the control. A slight, but significant decrease in the serum total protein concentration (TP) can be observed in a dose-dependent manner in both studies, although in one sex only in each study, i.e. in males of study A and in females of study B. An assessment of individual animal data revealed five animals with serum total protein values below 50 g/L. Four of them (animal Nos. 1, 6, 283 and 286) were in high dose groups (33% MON810) and one (animal No. 304) in the PR33D48 low dose (11% MON810). Normal serum total protein values are characterized by relatively low variability (coefficient of variation 5-8%). Normal serum total protein values for Wistar rats are in the range between 50 and 70 g/L (e.g. Boehm et al. 2007 , Wang et al. 2010 , Yamatoya et al. 2012 ). A decrease in serum protein can have various reasons, including chronic inflammation as related to carcinogenesis (cf. Ohki et al. 2012 ), and nephrotic syndrome (cf. Palanisamy et al. 2008 ). To further elucidate possible causes of the hypoproteinaemia observed in animal Nos. 1, 6, 283, 286 and 304, the histological slides of these animals should be re-assessed by a qualified pathologist with emphasis on inflammatory changes, in

GRACE Team Response



X


X


X



If the rats were suffering from a nephrotic syndrome, X oedema and histological alterations of the kidneys would (Considered become evident. In the experimental groups showing a in particular decrease in serum total protein levels neither oedema nor in a histological alterations of the kidneys were observed in comment male rats fed the 33% GMO diet in feeding trial A and in by female rats fed the 33% GMO diet in feeding trial B. Steinberg Furthermore, if protein synthesis were impaired, a certain [2015]) degree of liver parenchymal cell death would have become evident, but the histopathological analyses of the liver of male rats fed the 33% GMO diet in feeding trial A and in 43

No

58

59


GRACE Team Response

particular because questions are raised above with regard to the quality of the histopathological assessment. the Hungarian Competent authority, based on its scientific advisors, in connection with these 90 days studies, makes the following suggestions: - Male and female rats have different growing characteristics, metabolism and different hormonal system coupled with different sensitivity to diets/components. There is no scientific reason to assume that male and female animals would grow similarly. Therefore, a11 those data and results gained with the 90 day experiments which are significantly different between control and the GM diets should be taken into consideration and be reexamined, even if they are not uniform for males and females. - There is a great difference between observing or measuring some of the organs/organ histology. - Hungarian experts have suggested already at the start of the project that great care should be taken with the gastrointestinal tracl. The gut is the first line of contact between (GM) food and the body. In spite of this, no small/large intestinal data has been collected, except for some histological samples. No histological examinations have been performed on all the animals. - In a nutritional context, amino acid analysis is less important than measuring the NPU (net protein utilization) of the proteins consumed. Other questions and comments An extension of the recommended OECD test battery is also warranted according to the EFSA “Guidance on conducting repeated-dose 90-day oral toxicity study in rodents on whole food/feed” which clearly states that “Other endpoints should also be considered if there are indications that the whole food/feed may have effects on e.g. the cardiovascular, nervous, gastrointestinal tract or immune system. For instance, if the whole food/feed is expected to have an impact on the gut, then

female rats fed the 33% GMO diet in feeding trial B revealed no histological alterations. A careful control of the data reported by the Slovak Medical University was performed by two scientists of the consortium in Bratislava, and the vealuation showed that the data reported were the correct ones (1. comment). All organs in the control and the 33% GM maize-fed groups were examined at the macroscopic and microscopic level (i.e, the 2. comment does not apply). The gastrointestinal tract was anlyzed macroscopically as well as microscopically (i.e., the 3. comment does not apply). The measurement of the NPU was not foreseen in the study plans, but the growth curves and the weight gain revealed that there was no nutritional imbalance.




X

The mandatory regulatory inclusion of feeding trials into the risk assessment does not refer to triggers for conducting studies. Nevertheless, the project comparatively considers a set of additional and alternative approaches and endpoints.

44

No

60

61


the microbial flora should be investigated. Additional markers of potentially adverse nutritional and/or metabolic effects should be considered on a case-by-case basis, according to the available body of evidence and the type of whole food/feed under investigation . These “indications” are presented by Finamore et al. The given study batteries and designs do not allow any final conclusions on potential long term, reproductive or developmental effects, which are the most important and appropriate effects potentially induced by food and feedstuff as the “consumers” are exposed over their entire life span with the GM product. To publish and communicate the outcome and the data originating from the 90 day experiments resulted from GMO contaminated control material would be a wrong message to consumers, scientists, risk assessors and even for GMO developing companies when submitting new notifications. This message may be that the European Commission, EFSA and Member States' experts accept it as facts: - that maize production in Europe is contaminated and it is impossible to grow GMO free maize in Europe. Considering that the EU still maintains a zero tolerance policy as regards unauthorised GMO contamination in non GM sowing seeds, accepting any level of contamination in the maize varieties used for diet preparation as control reference would send a wrong signals; - that the scientific standards of the EU or the EFSA could be weakened when they finance or participate projects on risk assessment of GMOs, and at the same time strengthened when dealing with industry, or when assessing the results of research identifying negative health or environmental effects of GMOs.

GRACE Team Response




The tests are proposed and considered as being models to facilitate a thorough risk assessment for human (and animal) consumption. Regarding this the project shall discuss design and rationale for conducting feeding trials as well as alternatives. (no Q; statement). Given the sensitive detection methods and given the presence of approved GMOs in food/feed material in commercially available ingredients and grains traces of GMOs can hardly be excluded from the diets; GRACE tried to minimize admixtures by careful planning and contracting.

45

No


GRACE Team Response




1-year chronic feeding study 62

63

64

65

The GRACE project is expected to give policy advice, and therefore A careful control of the data reported by the Slovak this advice should be based on well-defined and high quality Medical University was performed by two scientists of the conducted studies. consortium in Bratislava, and the vealuation showed that the data reported were the correct ones (1. comment). All organs in the control and the 33% GM maize-fed groups were examined at the macroscopic and microscopic level (i.e, the 2. comment does not apply). The gastrointestinal tract was anlyzed macroscopically as well as microscopically (i.e., the 3. comment does not apply). The measurement of the NPU was not foreseen in the study plans, but the growth curves and the weight gain revealed that there was no nutritional imbalance. Industry feeding studies qualify the grain test substances with Not possible within the strict time frame imposed by the event-specific PCR and mycotoxin testing before approving its use European Commission. in whole feed feeding studies. If an applicant had a similar situation in one of the animal feedings studies meant to be submitted to EFSA for evaluation, the applicant would have stopped the study and produced more grain to ensure that mycotoxin contamination did not confound the results of the study. Based on the disclosed contaminant levels of fumonisins in a The level of fumonisin contamination corresponds to the control grain sample, and the apparent lack of data regarding NOAEL for fumonisins in rats. Nevertheless, sphinganine contaminant levels in the diets currently being fed to the animals and sphingosine levels in the blood of the rats will be in the chronic study, it is possible that the fumonisin determined, in order to be able to determine whether the contamination of the control diet in the chronic feeding study levels of fumonisin in the diets were toxicologically may preclude an appropriate comparison of data obtained from relevant or not. the animals being fed the test diets, therefore negatively impacting the study results and compromising the utility of the data for its intended purpose. If the outcome of contaminant analysis of the diets currently The level of fumonisin contamination corresponds to the being fed to the animals demonstrates fumonisin levels near or NOAEL for fumonisins in rats. Nevertheless, sphinganine above levels of concern, it would be prudent to discontinue this and sphingosine levels in the blood of the rats will be

X

NA

X

NA

X

NA

X

46

No


GRACE Team Response



study, started in Feb 2014, and focus on the production of uncontaminated grain for the 2-y study to start later this year. 66

67

68

determined, in order to be able to determine whether the levels of fumonisin in the diets were toxicologically relevant or not. As disclosed during the stakeholder consultation, there is intent A careful analysis of the contaminants in the diet batches is to produce 2 batches of diets for the ongoing chronic study. Based foreseen. on the existing information regarding contaminant levels in the control grain and uncertainties regarding contaminant levels in the diets currently being fed, it is advisable to re-evaluate the stored grain for contaminants, as levels can change depending on the conditions of storage. It is also advisable to evaluate the second diet batches for contaminants prior to feeding them to the test system. A primary reason given for the inclusion of reference groups fed The use of historical background data from the animal diets prepared using conventional maize grain in the subchronic supplier is highly problematic: we are intensely discussing and chronic studies is that the test facility where the studies are this point in the GRACE as well as in the G-TwYST being conducted does not have sufficient historical control data to consortium in an intense discussion between toxicologists provide an accurate evaluation of normal biologic variability of and statisticians. Although we are trying very hard to get the test system, an important component for contextualization the primary data from the Harlan database, in the end we and interpretation of observed statistical differences. In that will have to use the data generated by GRACE as the regard, the absence of 3 of 4 conventional reference groups from reference. the chronic study has already compromised the interpretability of the data obtained from this study. As indicated above with respect to data interpretation from the subchronic studies, it is recommended that historical background data be obtained from the animal supplier and used to supplement the paucity of available historical data at the test facility. As indicated during the stakeholder consultation, it is the position We principally agree with this comment, and all data will of the GRACE consortium that it is too early to draw conclusions be made available to the stakeholders. from the subchronic study, and the chronic study is being conducted, in part, to clarify observed differences from the subchronic data. In that regard, and based on the timing of the chronic study, data from the 3-month interim clinical pathology evaluation should be available for statistical analysis and review.


NA

X

NA

X

NA

X

47

No


69

It is therefore requested that the GRACE consortium complete the statistical analysis of this interim data and provide an update for interested stakeholders. Observed differences from the subchronic studies, if related to consumption of the test diets, should be reproducible within the chronic study at a comparable feeding interval. Consideration should be given to the differences in statistical power, as a lower number of animals (10/sex) are scheduled for interim sampling in the chronic study as compared with the number of animals that were sampled at the comparable interval in the subchronic studies (16/sex). Chronic study. I believe the OECD calls for more animals per group. You have 10 per group. That was a major criticism of the Seralini study and if memory serves, the numbers should be somewhere closer to 20, although it depends in part on the outcomes you are trying to measure. The Wistar Han is probably a better strain of rats to use compared to Sprague Dawly for chronic studies. Fumonisin contamination is reported for the test material used in the 1-year study. Clarification is needed on the actual material that showed contamination and the potential impact on the ongoing study.

70

71

Given the contamination of test material with fumosin, the one year study should be cancelled and maybe restarted, as it is apparent that the study will not yield results that could clarify the findings of the 90 day studies.

72

In case the study is continued, please elaborate on the necessary modifications in hypothesis, analysis and comparison with the findings of the two 90 day studies.

GRACE Team Response



Twenty animals per group are being used in the 1-year MON810 study.

NA

The level of fumonisin contamination corresponds to the NOAEL for fumonisins in rats. Nevertheless, sphinganine and sphingosine levels in the blood of the rats will be determined, in order to be able to determine whether the levels of fumonisin in the diets were toxicologically relevant or not. The level of fumonisin contamination corresponds to the NOAEL for fumonisins in rats. Nevertheless, sphinganine and sphingosine levels in the blood of the rats will be determined, in order to be able to determine whether the levels of fumonisin in the diets were toxicologically relevant or not. The contamination level as well as the result of the sphinganine and sphingosine levels in the blood of the rats will be taken into account when analyzing the findings of the 1-year feeding trial for itself and when comparing the

NA

X

NA

X

NA

X


X (the comment does not apply)

48

No

73


GRACE Team Response

results of the 90-day and the 1-year MON810 feeding trials. The one year long experiment has started before the GM and The level of fumonisin contamination corresponds to the control maize components of the diet were analysed for GM NOAEL for fumonisins in rats. The contamination level as contamination and other toxicants. Because some maize lines had well as the result of the sphinganine and sphingosine levels a high mycotoxin contamination, and this maize was used as part in the blood of the rats will be taken into account when of a diet, we cannot be sure if any of the effects identified were analyzing the findings of the 1-year feeding trial for itself caused by GM maize, or by mycotoxin contamination. Under such and when comparing the results of the 90-day and the 1conditions the careful work of the Bratislava group is being year MON810 feeding trials. Due to the strict time frame wasted. Therefore, we urge that the long term study should be imposed by the European Commission, no alternative finished without delay and a new study has to be started with the approach is possible. GM line, with new controls, where there are neither GMO, or mycotoxin, nor other contaminants present.


NA



X

Alternative approaches 74

We believe the GRACE team should bear in mind that these alternative methods have been included in the EU project as “exploratory” or “experimental” techniques; until such methods are fully validated they should not be considered for regulatory purposes.

75

The fact that the corn used in these studies was grown in unreplicated plots that were physically separated from one another means that the results are prone to bias. This is a particular concern for ‘omics-type assessments which can be very sensitive, and which may well detect differences that are a reflection of differences between the chosen fields rather than genuine genotype effects.

A substantial part of our GRACE activities should have focused on alternative methods to e.g. substitute animal experimentation (one important claim of the EU call). Therefore we aimed introducing suitable in vitro methods to archive new information about the potential and validity of such techniques. We are aware of the current limitations of such alternative methods on the route detecting unintended food/feed effects. However, as a potential outcome of the alternative approaches (‘omics’techniques and RRR) it might enable an objective judgement of their future value within the regulatory process. Plots were physically separated to minimize crosspollination between different varieties. However, all varieties were cultured in a very reduced area with overall the same type of soil and using the same cultural practices. Moreover, two pairs of MON810 and near-isogenic varieties were grown and analysed using ‘omics, together with up to 5 additional conventional varieties to capture

NA

NA

49

No

76


GRACE Team Response

variability associated to the specific commercial variety, agricultural aspects (including the chosen fields). As it is recognized by their specific agricultural characteristics and has been previously demonstrated using omics approaches, the differences between conventional varieties is much larger than those dependent on agricultural and environment aspects. Finally, growing replicate plots would very much complicate the work at the agricultural level, e.g. increasing the risk of adulteration, poor performance or failure. We believe that the interpretation of the longitudinal The longitudinal follow-up of animals is equivalent in the metabolomics, with regard to differentiating adverse vs. adaptive present 90-day study design to repeated analyses effects, is challenging. The animal model that is being used is not performed at different dates (about one per month in the to our knowledge validated. The biological range of detection of present case). It corresponds to the duplication of the last each metabolite in naïve animals is not known. The inclusion of day (91 day) observation performed prior to it. In the “reference compounds” would help in validating this alternative present study, no commercial cultivars were used because approach in order to detect true toxic effects. In addition, caution they cannot be really included in a more sophisticated should be taken when interpreting the metabolic differences experimental design with 4 controlled factors, that is observed in the present study, as no samples replicates have been gender, dose of GM-maize in diet, cultivar (GM and the included in order to compensate for the various bias that may respective isogenic conventional comparator) and date. influence the integrity of the harvested crop samples. It would be With our experimental design, it is possible to model relevant to use a similar field study design to what EFSA variations of macroscopic (growth rate, feed consumption, recommends for the comparative analysis of composition etc.), biochemical and metabolomic variables using the (minimum of 8 cultivation sites, minimum of 4 replicates/site, mixed-effect model which corresponds to a specific anova randomized block design). designed to organise a part of the variance to every animal for which there are repeated analyses. Indeed, the factor "dose" corresponding to the level of incorporation of GM maize in diet is a convenient way to check, whatever the cultivar considered (block effect), a possible impact of GM trait on metabolism and possibly on macroscopic variables such as growth rate modeled on the whole experimental follow-up period. This metabolonic research needs to be based a relative exploration, not on an absolute one. The




NA

50

No


GRACE Team Response




crop variability may be interesting to model at the animal model level (it is always a statistical canonical problem consisting to use the animal model as a valuable "instrument" to phenotype at a more functional level the maize grain composition and its inherent variability), but at what cost and for what additional conclusion ? This should be envisaged latter on in dedicated multicentric experiments and should be based on a formal statistical meta-analysis modeling approach. 77

78

Omics More on metabonomics. Does anyone have a dataset that has recorded the variation in rats fed different diets? What happens in the “untargeted” screen if diets vary substantially in specific carbohydrates, fatty acids, total protein, digestible fiber, specific vitamins? I really do not think you can possibly learn much from this study, except some very preliminary data on natural variation under a very restricted set of variables. It seems like a missed opportunity. A lot of money is being spent and animals sacrificed and yet so much more important information might have been gained by altering the study plan to include a few intentionally different diets. Longitudinal metabolomics studies Cry1ab degradation: Regarding the longitudinal metabolomics study presentation and related discussions of Cry1ab degradation, EuropaBio would like clarification of the GRACE Project’s results to-date. Several studies by independent researchers have been published which examine the fate of Cry gene and proteins following dietary ingestion. They repeatedly find that the intact protein and gene are not detected in the plasma or systemic tissues of the test subjects (Chowdhury et al., 2003 ; Chowdhury et al., 2004 ; Walsh et al., 2011 ; Walsh et al., 2012 ; Sieradzski et al., 2013 ; Świątkiewicz et al., 2013 ). Thus, the protein was not absorbed from the gut intact and could not act systemically.

Very little natural variation may be observed in practice as all animals have received the same diet. This may lead to observed differences between the groups that are not directly related to the genetic modification, but rather to (small) changes in other aspects.

NA

Comment actually not requiring an answer from the GRACE team. We agree that in certain cases the pig may be a more relevant assessment model than the rat, but not in the case of wanting to determine if the genetically modified maize MON810 leads to chronic toxicity.

NA

X

51

No

79


Moreover, many of these studies were performed in pigs (Walsh et al., 201114; Walsh et al., 201215; Sieradzski et al., 201316; Świątkiewicz et al., 201317). This is a noteworthy observation, because the digestive anatomy and physiology of pigs are considered similar to humans by subject matter experts (Swindle et al., 2012 ; Kararli, 1995 ). Thus, the pig may be a more relevant assessment model than the rat. Antibody measurements against the GM protein. Have you included a design to test antibody production to other dietary proteins? The Cry1A protein is expressed at a very low level in maize. The probability of generating antibodies against it is extremely remote. But you also do not have a standard to measure it against. You should be measuring antibody responses to a number of dietary proteins of different concentrations and characteristics to understand more about the generation of antibodies to dietary proteins. Most humans have some antibodies of different isotypes against some dietary proteins. Although I think it will generally be to more highly abundant proteins. So, is this a waste of time and money since you don’t have a basic understanding of antibody responses to other dietary proteins?

GRACE Team Response


It was planned to determine total IgE, IgG and IgM levels as well as specific anti-Cry1Ab antibody levels, and this has been done in the meantime. Moreover, a number of immunotoxicity related endpoints have been measured in the 90-day longitudinal study, but the determination of "antibody responses to a number of dietary proteins" was not included in the study design and has not been performed.

NA

Two NK603 dose levels will be tested, while the strongly criticized MON810 study with 1 dose level will not be performed.

NA

The limited animal housing capacity and the financial frame do not allow to include one or the other conventional reference group.

NA



X

G-TwYST and next steps 80

81

Several participants, including Member States authorities raised concerns about the interpretation of the intended 2-y studies missing different test concentrations. Therefore the scientific value and solidity of such a study is limited and open for public criticism and requests for repetition and other studies. The long term study plans, presented during the stakeholder meeting, did not include any conventional reference groups. This remains a major concern to EuropaBio for the reasons previously stated regarding the importance of historical control data in determining the normal range of variability in the animal model,

X

X

52

No

82

83

84


GRACE Team Response

and thus the biological relevance of any differences detected in a study utilizing that animal model. We encourage the G-TwYST to reconsider the study plans and Due to the limited animal housing capacity and the figure out a way to balance the number of groups in the studies in financial frame, the 2-year study will not include further order to compensate for the lack of historical background data. groups. We have left out the 2-year MON810 study and will include a second 90-day feeding trial testing the 50% GM maize inclusion level, so that we will have an enlarged background database at the Slovak Medical University. We would advise GRACE to concentrate on the NK603 studies and We agree with this comment and have left out the 2-year exclude the carcinogenicity study planned with MON810. By MON810 study. focusing on NK603, we believe the consortium would be able to save the resources and funding needed to complement the NK603 studies (90-d feeding study and combine chronic/carcinogenicity study) with appropriate historical background data. As regarding future studies, we recommend that, when carrying The procedure described by the stakeholder is the one out feeding experiments/projects with GMOs, the first step actually being performed by G-TwYST. should be to check all crops and feed for GMO-, and other contaminations, including mycotoxins. Only crops of high quality and reaching the highest scientific standards/criteria should be used for diet preparation. The prepared diets should be checked for contamination and quality again.



NA

X

NA

X

NA

X


Supporting documents (distributed before the workshop) 85

86

Ad Table 3: The relevance of the analysis of microorganisms after irradiation of the samples is not clear. The authors indicate in all instances colony forming unit counts below 10 CFU/g fresh weights. The exact numbers would be easily determined by agar plate counting. The criteria for differentiating “Total Coliforms” (= Enterobacteriaceae), “Enterobacteriaceae”, “E. coli” (= Enterobacteriaceae) and “Salmonellae” (= Enterobacteriaceae) are not evident. The same is true for “Yeasts” and “Molds”. Please explain. Ad Table 4: The explanations for indexes 1 – 6 are missing.

The indication that in all instances colony forming units are below 10 CFU/g fresh weight shows that there is no biologically relevant contamination of the diets with microorganism, so that no further analyses were foreseen and are needed.

X

(no Q; statement)

53

Appendix 1: Agenda 2nd GRACE Stakeholder Workshop on Animal Feeding Studies and In Vitro Studies in GMO Risk Assessment Agenda 19 May 2014 11:30

12:30

Registration Possibility for light lunch at canteen (at own expense) INTRODUCTION Welcome Patrick Rüdelsheim, Perseus, Ghent, Belgium Update on the GRACE project by the Coordinator Joachim Schiemann, Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany Stakeholder and User Involvement in GRACE/ Introduction to the Workshop Armin Spök, IFZ-Inter-University Research Centre for Technology, Work and Culture, Graz, Austria Implementation of the feeding studies: requirements and constraints Ralf Wilhelm, Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany Discussion

13:45

90-DAY FEEDING STUDIES – DESIGN AND ANALYSIS Chair: Armin Spök General design and analysis Pablo Steinberg, University of Veterinary Medicine, Hannover, Germany Maize cultivation Pere Puigdomènech, Center for Research in Agricultural Genomics (CRAG) Barcelona, Spain Discussion

15:00

15:30

Coffee break

Analysis of plant material and diets Gijs Kleter, RIKILT-Institute of Food Safety, Wageningen University and Research Centre, The Netherlands Test facility, periodic observations and necropsy Dagmar Zeljenková, Department of Toxicology, Slovak Medical University, Bratislava, Slovakia Haematology and biochemistry Jana Tulinska, Department of Toxicology, Slovak Medical University, Bratislava, Slovakia Histopathology Pablo Steinberg, University of Veterinary Medicine, Hannover, Germany Discussion

18:00

Adjourn Day 1

Agenda 20 May 2014 08:30

08:45

Recapitulation day 1/ launch day 2 Patrick Rüdelsheim, Perseus, Ghent, Belgium ALTERNATIVE APPROACHES Chair: Gijs Kleter In vitro studies Ralf Einspanier, Freie Universität, Berlin, Germany Omics studies on plant material Esther Kok, RIKILT-Institute of Food Safety, Wageningen University and Research Centre, The Netherlands

2

Longitudinal metabolomic studies Karine Adel-Patient, Unité INRA d'Immuno-Allergie Alimentaire, Gif-sur-Yvette, France Discussion 10:15

Coffee break

10:45

PROSPECTS AND NETWORKING Chair: Joachim Schiemann Further studies Ralf Wilhelm, Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany CADIMA Christian Kohl, Institute for Biosafety in Plant Biotechnology, Julius Kühn Institut, Quedlinburg, Germany G-TwYST Pablo Steinberg, University of Veterinary Medicine, Hannover, Germany Discussion

12:15

GENERAL DISCUSSION Chair: Armin Spök General discussion Next steps and closing remarks

13:45

End of workshop Possibility for light lunch at canteen (at own expense)

3

Appendix 2: List of Participants GRACE Team Members Gloria Adduci, IFZ - Interuniversitaeres Forschungszentrum für Technik, Arbeit und Kultur, Austria Karine Adel-Patient, Unité INRA d'Immuno-Allergie Alimentaire, France Ralf Einspanier, Freie Universität Berlin, Germany Gijs Kleter , RIKILT-Institute of Food Safety, Wageningen University and Research Centre, The Netherlands Christian Kohl, Julius Kühn-Institut (JKI), Bundesforschungsinstitut für Kulturpflanzen, Germany Esther Kok, RIKILT-Institute of Food Safety, Wageningen University and Research Centre, The Netherlands Klaus Minol, Genius GmbH, Germany Pere Puigdomènech, Center for Research in Agricultural Genomics (CRAG), Spain Patrick Rüdelsheim, Perseus bvba, Belgium Joachim Schiemann, Julius Kühn-Institut (JKI) Bundesforschungsinstitut für Kulturpflanzen, Germany Kerstin Schmidt, BioMath GmbH, Applied Statistics and Informatics in Life Sciences, Germany Greet Smets, Perseus bvba, Belgium Armin Spök, IFZ - Interuniversitaeres Forschungszentrum für Technik, Arbeit und Kultur, Austria Pablo Steinberg, University of Veterinary Medicine Hannover, Germany Jana Tulinska, Slovak Medical University, Department of Toxicology, Slovakia Justus Wesseler, Wageningen University, Agricultural Economics and Rural Policy, The Netherlands Ralf Wilhelm, Julius Kühn-Institut (JKI), Bundesforschungsinstitut für Kulturpflanzen, Germany Dagmar Zeljenkova, Slovak Medical University, Department of Toxicology, Slovakia Academia/Research Didier Breyer, Scientific Institute of Public Health, Belgium René Custers, VIB, Belgium Adinda De Schrijver, Scientific Institute of Public Health, Belgium David Demortain, INRA SENS/UPEM, France Franck Fourès, Agence nationale de sécurité sanitaire, France Lieve Gheysen, Ghent university, Department Molecular Biotechnology, Belgium Wim Grunewald, VIB, Expert scientist Plant biotechnology, Belgium Joël Guillemain, HCB Scientific committee, France An Hagenaars, University of Antwerp, Zebrafishlab, Belgium Dries Knapen, University of Antwerp, Zebrafishlab, Belgium Harry Kuiper, formerly RIKILT WUR, Institute of Food Safety, The Nederlands Peadar Lawlor, Teagasc, Pig Development Department, Animal and Grassland Research and Innovation Centre, Irland Jasmien Maes, Ghent University, Department Molecular Biotechnology, Belgium Andreas Mitzschke, Maastricht University, Department of Technology and Society Studies, The Nederlands Simona Panzacchi, Centro di Ricerca sul Cancro Cesare Maltoni Istituto Ramazzini, Italy Ad Peijnenburg, RIKILT Wageningen University and Research Centre Toxicology Group, The Nederlands Ine Pertry, Institute of Plant Biotechnology Outreach (IPBO) Ghent University, Belgium Eline Rademakers, Expert Novel Foods and GMO's | Animals, Plants and Foodstuffs | Food, Feed and Other Consumption Products, Belgium Bernard Salles, TOXALIM - UMR 1331 INRA/INP/UPS, France Frederic Schorsch, European Society of Toxicologic Pathology, Belgium Piet van der Meer, University of Ghent/Public Research and Regulation Initiative (PRRI), Belgium Frank van der Wilk, COGEM, The Nederlands

Competent Authorities Hans Christer Andersson, National Food Agency, Sweden, Sweden Arnaud Apoteker, The Greens/ EFA Group in the European Parliament, Belgium Zsuzsanna Bardocz , Biodiversity and Gene Conservation Unit, Department for Strategy, Hungary Doris Bühler, FOAG (Swiss Federal Office for Agriculture), Switzerland Omar del Río Fernandez, Ministry of Agriculture, Food and Environment, Spain Jens Hoegel, European Commission DG Research & Innovation, D2 Advanced Manufacturing Systems and Biotechnologies, Belgium Hajnalka Homoki, Ministry of Rural Development of Hungary, Hungary Kaia Kantorska, European Commission, DG Health and Consumers, Unit E1 - Biotechnology, Belgium Joachim Kreysa , DG JRC, Institute of Health and Consumer protection, Molecular Biology & Genomics, Belgium Anna Lanzoni, EFSA, Italy Volker Matzeit, Federal Ministry of Food and Agriculture, Division 222 - New Technologies , Germany Françoise Mori, Permanent Representation of Luxembourg to the European Union, Belgium Dominique Perreaux, Belgian Biosafety Advisory Council, Belgium Annette Pöting, Federal Institute for Risk Assessment, Unit Food Toxicology, Department Food Safety, Germany Andrea Scheepers, Bundesamt für Verbraucherschutz und Lebensmittelsicherheit, Abteilung Gentechnik, Germany Gianpaolo Suriano, European Commission, DG Research & Innovation F2 –Biotechnology, Belgium Dietmar Vybiral, Federal Ministry of Health, Austria Birgit Winkel, Federal Agency for Nature conservation, Germany Civil Society Organisations Andreas Bauer-Panskus, Testbiotech, epigen Wissenschafts- und Projektbüro, Germany Kirsty Reid, Eurogroup for Animals, Belgium Kristina Wagner, Eurogroup for Animals, Germany Industry Roland Buesen, BASF Delphine Carron, EuropaBio, Belgium Rocío Fernández Cantón , Monsanto Violeta Georgieva, EuropaBio, Belgium Jean-Baptiste Rascle, Bayer Suzy Renckens, Syngenta, Belgium Jason Roper, DuPont, USA Feyza Selcuk, Monsanto Sabine Storck-Weyhermueller, Syngenta Crop Protection, Switzerland Andrew Tommey, Pioneer Overseas Corporation, Belgium Guido Velten, Bayer Keith Ward, Syngenta Irina Wenderoth, BASF

2

Appendix 3: Workshop Presentations

Appendix 3 Slide Presentations On the following pages the slide presentations are included in the order as they were presented at the stakeholder consultation workshop.

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

Gloria1801

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

Appendix 4: Comments Received in Writing Written comments were received from the following organisations or individuals resp.: EFSA EuropaBio, Belgium European Society of Toxicologic Pathology/Bayer, France Federal Agency for Nature Conservation (BfN), Germany Hungarian Ministry of Agriculture, Hungary Istituto Ramazzini, Italy Österreichische Agentur für Gesundheit und Ernährungssicherheit GmbH, AGES, Austria Public Research and Regulation Initiative (PRRI), The Netherlands Richard E. Goodman, University of Nebraska, Lincoln, USA Testbiotech, Germany

Istituto Ramazzini, Italy

I really appreciated the job you did for this project and also the dissemination of the information regarding it. I have only a technical question for you, regarding the periodical observation on rats and its statistical analysis. In your excel file, in data used for statistics there is the gain of body weight for each cage per day. Why did you considered not the week but the day? Is there some reference for this? Or it is because for the long term study you have to evaluate the food efficiency ratio and you probably will do this ratio on day and not week? Why did you decide to not plot the gain? Where is the statistic report for feed consumption (you mentioned the analysis in material and method page 9)? Thank you for the attention Simona Panzacchi Dott.ssa Simona Panzacchi Centro di Ricerca sul Cancro Cesare Maltoni Istituto Ramazzini , Bentivoglio

1

Federal Agency for Nature Conservation (BfN), Germany Bonn, 18 June 2014 „GMO Risk Assessment and Communication of Evidence“(GRACE) Remarks of the Federal Agency for Nature Conservation (BfN), Germany, concerning the “Stakeholder Consultation on Animal Feeding Studies and In-Vitro Studies in GMO Risk Assessment that took place from 19th – 20th May 2014 in Brussels Maize cultivation The cultivation of the GMP and the near-isogenic line took place in a certain distance to each other. The reason for this should be given and the possible influence on the composition of the plants should be discussed. It could lead to a higher variability between the lines because of different conditions and therefore cover effects caused by the genetic modification. Diet As it is not expected that GMO will be highly toxic, the GMO-content should be as high as possible. A content of 33% is not the highest possible concentration without putting an extra stress on the rats and is therefore not optimal to find effects. Also the presence of GMP in control maize varieties can mask effects caused by the genetic modification and therefore the relevance of the outcome of the study can be questionable.

2

Public Research and Regulation Initiative (PRRI), The Netherlands GRACE Stakeholder Consultation 19 June 2014, Brussels.

Given its assignment and composition, the GRACE project consortium is in a unique position to help remind/clarify a number of important characteristics of Environmental risk assessment (ERA) and food/feed safety assessment (FSA) of GM crops in the context of regulatory decision making:

1) The principles and methodology of ERA and FSA are based on internationally developed approaches such as Annex III of the Cartagena Protocol on Biosafety (CPB) and guidelines of the Codex Alimentarius.

2) While ERA and FSA have different assessment endpoints and characteristics, they share that they need to be conducted in a manner that is: a. scientifically sound and transparent b. well structured, stepwise c. tiered and hypothesis driven, i.e. early stages in the assessment of a specific assessment endpoint (e.g. toxicity) indicate whether and what additional information or testing may be required, i.e. those next steps have a scientific rationale ( ‘hypothesis driven’ ). d. Additional data should only be asked when one knows what to do with the answer e. comparative - identified potential adverse effects are considered in the context of the risks or effects posed by the non-modified recipients and their use.

3

Österreichische Agentur für Gesundheit und Ernährungssicherheit GmbH, AGES Comment on the “90-day oral toxicity studies on two genetically modified maize nd

MON810 varieties in Wistar Han RCC rats” as presented on the 2 Workshop of the GRACE consortium (19. – 20.5.2014)

Stakeholder

1. Major remarks a) By inclusion of additional conventional maize varieties into the feeding scheme naturally occurring variability between conventional maize varieties and their effect on the consumer can be determined. However, including too many collectives, clusters, controls, and in this specific case even verum varieties increases statistical variances and reduces the statistical power of the analysis. This approach may obscure significant effects induced by the genetically modified feed constituent. An example is the significant and even dose-dependent diminution of the pancreas weight with male rats in feeding trial A. This effect is completely eliminated if the data of both feeding trials A and B are merged and statistically evaluated in a combined fashion. We recommend putting the focus on results obtained from animals fed with the GM variety compared to the study group receiving the near-isogenic non-GM comparator containing feed. b) As only minute concentrations of the active principle (i.e. the transgenic protein) are present in the diet a very robust and precise study design with a minimum of biological deviations and bias is mandatory to discover differences in the effects mediated by the GM maize containing feed compared to the study group exposed only to the near-isogenic non-GM comparator containing feed. c) To detect any effects induced by genetically modified maize containing feed during this short period of exposure (i.e. 90 days) the tested endpoints have to be extended to include also immunological parameters like B- and T-cell subpopulations, the analysis of changes in membrane receptors of theses lymphocytes and the quantification of cytokines. Finamore et al. (2008) evaluated the gut and peripheral immune response to genetically modified (GM) maize in mice in vulnerable conditions.(1) Weaning and old mice were fed a diet containing MON810 or its near-isogenic control maize or a pellet diet containing a GM-free maize for 30 and 90 days. The immunophenotype of intestinal intraepithelial, spleen and blood lymphocytes of control maize fed mice was similar to that of pellet fed mice. As compared to control maize, MON810 maize induced alterations in the percentage of T and B cells and of CD4+, CD8+, γδT, and R T subpopulations of weaning and old mice fed for 30 or 90 days, respectively, at gut and peripheral sites. An increase of serum IL-6, IL-13, IL-12p70, and MIP-1 after MON810 administration was also observed. The results were indicative for the importance of the gut and peripheral immune response to GM crop ingestion as well as the age of the consumer in the GMO safety evaluation. In the present study no specific measurements/analyses were performed to validate the results of Finnamore et al. d) An extension of the recommended OECD test battery is also warranted according to the EFSA “Guidance on conducting repeated-dose 90-day oral toxicity study in rodents on whole food/feed” which clearly states that “Other endpoints should also be considered if there are indications that the whole food/feed may have effects on e.g. the cardiovascular, nervous, gastrointestinal tract or immune system. For instance, if the whole food/feed is expected to have an impact on the gut, then the microbial flora should be investigated. Additional markers of potentially adverse nutritional and/or metabolic effects should be considered on a case-by-case basis, according to the available body of evidence and the type of whole food/feed under investigation.”(2) These “indications” are presented by Finamore et al.(1) e) The given study batteries and designs do not allow any final conclusions on potential long term, reproductive or developmental effects, which are the most important and appropriate effects potentially induced by food and feedstuff as the “consumers” are exposed over their entire life span with the GM product.

4

f) A GM maize concentration of 2,6% in the conventional control 1 of study B is inacceptable as non-GM comparator.

2. Minor remarks a) Ad Table 3: The relevance of the analysis of microorganisms after irradiation of the samples is not clear. The authors indicate in all instances colony forming unit counts below 10 CFU/g fresh weight. The exact numbers would be easily determined by agar plate counting. The criteria for differentiating “Total Coliforms” (= Enterobacteriaceae), “Enterobacteriaceae”, “E. coli” (= Enterobacteriaceae) and “Salmonellae” (= Enterobacteriaceae) are not evident. The same is true for “Yeasts” and “Molds”. Please explain. b) Ad Table 4: The explanations for indexes 1 – 6 are missing.

References 1

2

Finamore A, Roselli M, Britti S, Monastra G, Ambra R, Turrini A, Mengheri E. 2008. Intestinal and peripheral immune response to MON810 maize ingestion in weaning and old mice. J Agric Food Chem. EFSA. 2011. Guidance on conducting repeated-dose 90-day oral toxicity study in rodents on whole food/feed. The EFSA Journal 9:2438 - 2459.

5

GRACE stakeholder consultation 19/20 May 2014 EuropaBio written comments EuropaBio is the European Association for Bio-industries and was created in 1996. Our mission is to create an innovative and dynamic biotechnology-based industry in Europe. We acknowledge and appreciate the fact that the GRACE consortium is conducting an extensive amount of work in the framework of the FP7 project, while maintaining transparency and actively engaging stakeholders. We would also like to thank the GRACE consortium for the informative stakeholder consultation organised on 19-20 May in Brussels and for the opportunity to provide written comments. Please find below the EuropaBio comments for your consideration. They are structured following the different GRACE work streams that were presented during the consultation. 1. General Comments Policy advice As indicated in the GRACE consultation report (March 2013)1 this project is expected to provide policy advice: ‘The aim is to evaluate the EFSA guidelines in order to provide a scientific basis for policy development and to define gaps in RA’. As communicated in previous consultations, EuropaBio is concerned about the fact that the GRACE project seems to be ad-hoc tailored to respond to policy-making demands and would like to emphasize that the aim of the Research Framework Programme funding should be to support research and innovation with the final end of reaching the goals of growth, competitiveness and employment. Study rationale As raised during previous consultations by EuropaBio and other participants, the rationale for conducting the studies and the main objective of the project was not considered in sufficient detail. We would like to emphasize that the outcome of the study should include recommendations on how to conduct 90-day feeding studies. Following more than 150 studies in animals with food produced from GM plants, which were substantially equivalent to their non-GM comparator, no unintended biologically relevant effects related to their safety as food were detected (Flachowsky 20122; Snell et al., 20123). Use of data for the CADIMA database project EuropaBio will consider and discuss internally if animal feeding data submitted in applications can be used for the CADIMA database project by GRACE for systematic review process. Submission of 90-day feeding studies in previous applications EuropaBio would like to clarify that 90-day feeding studies have been submitted by industry in the past, without a trigger but for expediency during the regulatory process as some Member States were asking for it in any case. However, evaluation of the voting behavior by Member States reveals that they vote the same way on product approvals regardless whether 90-day rat studies have been provided or not. EuropaBio members are of the opinion that when substantial equivalence (i.e., no biologically relevant difference) is shown between the GM crop and its conventional counterpart, there is negligible value in the performance of 6

animal feeding studies with rodents to assess food safety. This position is supported by international risk assessment experts who have concluded that whole food feeding studies are unnecessary on GM crops that have already been demonstrated to be substantially equivalent to their conventional comparators by the molecular, compositional, phenotypic, and agronomic analyses (EFSA, 20114; Kuiper, 20125). In addition, from an animal welfare perspective as well as the 3Rs commitment to replace, refine, and reduce in accordance with the European Commission’s Directive 2010/63/EU6, the use of animals in scientific research should be minimized. International quality standards Although the animal feeding studies are being/have been conducted according to the principles of Good Laboratory Practice (GLP), the generation of the plant material by the GRACE project does not seem to follow these principles. We would like to highlight the fact that this is not consistent with industry practice. According to the EFSA guidance and EU Commission regulations, toxicological studies have to fully comply with the principles of GLP or these will not be considered for scientific evaluation. Studies, other than toxicological studies, also have to comply with the principles of GLP or be conducted by organizations accredited with ISO standards. Given the purpose of the GRACE Project, and if policy decisions will be taken based on results from studies conducted under the GRACE project, the same level of scrutiny should be followed for the studies carried out by the GRACE Project as Industry would receive for one of its studies. Historical Background Data According to the 2011 EFSA Food/Feed guidance and Implementing Regulation 503/2013, the use of historical background data to determine the natural variation of test parameters is highly advisable. In the case of the 90-day studies, the chosen testing facility did not have any relevant historical data, so the GRACE team included two conventional controls that create a limited historical data set. In addition, we would like to highlight the fact that, in the case of the long term studies discussed during the stakeholder consultation, no conventional controls will be included (except in the case of the 1y study where one conventional control is included). A more robust historical background data set for the endpoints under investigation on the 90-day study is available from the animal supplier7. We encourage GRACE to use this third party data to assist in interpreting the biological relevance of differences observed between treatment groups. 2. Results 90-day studies The GRACE consortium re-iterated that it is too early to draw any conclusions from the results that would impact risk assessment for GM crops. We would like to emphasize that the main objective stated by the GRACE project is to provide further guidance on how to conduct a 90-day feeding study with GM material and not to open the discussion on the safety of GM (MON810).

Interpretation of existing 90d studies The fact that, for certain endpoints, the order of assessment was systematic, with replicates of each entry being assessed in groups of four rather than being fully randomized, has the potential to introduce bias which could lead to apparent differences between genotypes that

7

are in fact caused by temporal or other, non-treatment-related trends. This may explain why certain differences were apparently detected in Study 1 but not Study 2, and vice versa. Design and conduct of future 90d studies The two 90d studies both had sample sizes of 16 animals per group, which was chosen on the basis of achieving a certain probability of detecting a standardised effect size equivalent to one standard deviation. A key drawback of this approach is that standardized effect sizes take no account of toxicological relevance. An extensive review carried out by EuropaBio members in 2012 led to the conclusion that a sample size of 10-12 animals per group was sufficient to provide high power to detect toxicologically relevant effects across a range of key endpoints. A sample size of 10-12 animals per group therefore represents a more appropriate balance between power and ethical considerations regarding the use of animals in research. It is important to highlight that study plan parameters for 90-day feeding studies in EU studies might impact industry’s studies submitted in many other countries outside EU and set new standards. An example of this is the inclusion of pancreas weights on the GRACE 90-day studies. Pancreas weights are not required by OECD TG 408, and are not among the organ weights considered to contribute the most value to safety testing conducted with foods (Michael et al., 2007). Thus, the reason for including them in the current studies has not been sufficiently explained, and in the absence of a scientific rationale it should not create a new regulatory requirement. 3. 1 year chronic toxicity study Study design - The GRACE project is expected to give policy advice, and therefore this advice should be based on well-defined and high quality conducted studies. - Industry feeding studies qualify the grain test substances with event-specific PCR and mycotoxin testing before approving its use in whole feed feeding studies. If an applicant had a similar situation in one of the animal feedings studies meant to be submitted to EFSA for evaluation, the applicant would have stopped the study and produced more grain to ensure that mycotoxin contamination did not confound the results of the study. - Based on the disclosed contaminant levels of fumonisins in a control grain sample, and the apparent lack of data regarding contaminant levels in the diets currently being fed to the animals in the chronic study, it is possible that the fumonisin contamination of the control diet in the chronic feeding study may preclude an appropriate comparison of data obtained from the animals being fed the test diets, therefore negatively impacting the study results and compromising the utility of the data for its intended purpose. - If the outcome of contaminant analysis of the diets currently being fed to the animals demonstrates fumonisin levels near or above levels of concern, it would be prudent to discontinue this study, started in Feb 2014, and focus on the production of uncontaminated grain for the 2-y study to start later this year. - As disclosed during the stakeholder consultation, there is intent to produce 2 batches of diets for the ongoing chronic study. Based on the existing information regarding contaminant levels in the control grain and uncertainties regarding contaminant levels in the diets currently being fed, it is advisable to re-evaluate the stored grain for contaminants, as levels can change depending on the conditions of storage. It is also advisable to evaluate the second diet batches for contaminants prior to feeding them to the test system. - A primary reason given for the inclusion of reference groups fed diets prepared using conventional maize grain in the subchronic and chronic studies is that the test facility where the studies are being conducted does not have sufficient historical control data to provide an

8

accurate evaluation of normal biologic variability of the test system, an important component for contextualization and interpretation of observed statistical differences. In that regard, the absence of 3 of 4 conventional reference groups from the chronic study has already compromised the interpretability of the data obtained from this study. As indicated above with respect to data interpretation from the subchronic studies, it is recommended that historical background data be obtained from the animal supplier and used to supplement the paucity of available historical data at the test facility. - As indicated during the stakeholder consultation, it is the position of the GRACE consortium that it is too early to draw conclusions from the subchronic study, and the chronic study is being conducted, in part, to clarify observed differences from the subchronic data. In that regard, and based on the timing of the chronic study, data from the 3-month interim clinical pathology evaluation should be available for statistical analysis and review. It is therefore requested that the GRACE consortium complete the statistical analysis of this interim data and provide an update for interested stakeholders. Observed differences from the subchronic studies, if related to consumption of the test diets, should be reproducible within the chronic study at a comparable feeding interval. Consideration should be given to the differences in statistical power, as a lower number of animals (10/sex) are scheduled for interim sampling in the chronic study as compared with the number of animals that were sampled at the comparable interval in the subchronic studies (16/sex). 4. Alternative approaches - We believe the GRACE team should bear in mind that these alternative methods have been included in the EU project as “exploratory” or “experimental” techniques; until such methods are fully validated they should not be considered for regulatory purposes. Interpretation of alternative approaches - The fact that the corn used in these studies was grown in unreplicated plots that were physically separated from one another means that the results are prone to bias. This is a particular concern for ‘omics-type assessments which can be very sensitive, and which may well detect differences that are a reflection of differences between the chosen fields rather than genuine genotype effects. - We believe that the interpretation of the longitudinal metabolomics, with regard to differentiating adverse vs. adaptive effects, is challenging. The animal model that is being used is not to our knowledge validated. The biological range of detection of each metabolite in naïve animals is not known. The inclusion of “reference compounds” would help in validating this alternative approach in order to detect true toxic effects. In addition, caution should be taken when interpreting the metabolic differences observed in the present study, as no samples replicates have been included in order to compensate for the various bias that may influence the integrity of the harvested crop samples. It would be relevant to use a similar field study design to what EFSA recommends for the comparative analysis of composition (minimum of 8 cultivation sites, minimum of 4 replicates/site, randomized block design).

Cry1ab degradation Regarding the longitudinal metabolomics study presentation and related discussions of Cry1ab degradation, EuropaBio would like clarification of the GRACE Project’s results todate. Several studies by independent researchers have been published which examine the fate of Cry gene and proteins following dietary ingestion. They repeatedly find that the intact protein and gene are not detected in the plasma or systemic tissues of the test subjects

9

(Chowdhury et al., 20038; Chowdhury et al., 20049; Walsh et al., 201110; Walsh et al., 201211; Sieradzski et al., 201312; Świątkiewicz et al., 201313). Thus, the protein was not absorbed from the gut intact and could not act systemically. Moreover, many of these studies were performed in pigs (Walsh et al., 201110; Walsh et al., 201211; Sieradzski et al., 201312; Świątkiewicz et al., 201313). This is a noteworthy observation, because the digestive anatomy and physiology of pigs are considered similar to humans by subject matter experts (Swindle et al., 201214; Kararli, 199515). Thus, the pig may be a more relevant assessment model than the rat. 5. G-TwYST Interpretation of results - Several participants, including MS authorities raised concerns about the interpretation of the intended 2-y studies missing different test concentrations. Therefore the scientific value and solidity of such a study is limited and open for public criticism and requests for repetition and other studies. - The long term study plans, presented during the stakeholder meeting, did not include any conventional reference groups. This remains a major concern to EuropaBio for the reasons previously stated regarding the importance of historical control data in determining the normal range of variability in the animal model, and thus the biological relevance of any differences detected in a study utilizing that animal model. - We encourage the G-TwYST to reconsider the study plans and figure out a way to balance the number of groups in the studies in order to compensate for the lack of historical background data. - We would advise GRACE to concentrate on the NK603 studies and exclude the carcinogenicity study planned with MON810. By focusing on NK603, we believe the consortium would be able to save the resources and funding needed to complement the NK603 studies (90-d feeding study and combine chronic/carcinogenicity study) with appropriate historical background data. References 1 http://www.grace-fp7.eu/sites/default/files/block-files/Consultation%20Report.pdf 2 Flachowsky 2012, GMSAFOOD, report form Conference in Vienna 2012 (page 12) http://www.gmsafoodproject.eu/Conference/MagazineGMSAfood_webQ.pdf 3 http://www.sciencedirect.com/science/article/pii/S0278691511006399 4 EFSA Guidance for risk assessment of food and feed from genetically modified plants (2011). EFSA J. 9(5):2 150. http://www.efsa.europa.eu/en/efsajournal/doc/2150.pdf 5 Kuiper HA, Kok EJ, and Davies HV (2013): New EU legislation for risk assessment of GM food: no scientific justification for mandatory animal feeding trials. Plant Biotechnology Journal 1, 1–4. 6 http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32010L0063:EN:NOT 7http://www.harlan.com/products_and_services/research_models_and_services/research_models/wi star_han_outbred_rat/rcchanwist/italy.hl 8 Chowdhury EH et al. (2003) Detection of corn intrinsic and recombinant DNA fragments and Cry1Ab protein in the gastrointestinal contents of pigs fed genetically modified corn Bt11. J Anim Sci. Oct;81(10):2546-51. http://www.animal-science.org/content/81/10/2546.full.pdf 9 Chowdhury EH et al. (2004) Fate of maize intrinsic and recombinant genes in calves fed genetically modified maize Bt11.J Food Prot. Feb;67(2):365-70. http://www.ask-force.org/web/Bt1/ChowdhuryFate-Maize-Intrinsic-2004.pdf 10 Walsh MC, Buzoianu SG, Gardiner GE, Rea MC, Gelencsér E, Jánosi A, Epstein MM, Ross RP, Lawlor PG (2011). Fate of transgenic DNA from orally administered Bt MON810 maize and effects on immune response and growth in pigs. PLoS One;6(11) http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0027177

10

11 Walsh MC, Buzoianu SG, Rea MC, O'Donovan O, Gelencsér E, Ujhelyi G, Ross RP, Gardiner GE, Lawlor PG (2012) Effects of feeding Bt MON810 maize to pigs for 110 days on peripheral immune response and digestive fate of the cry1Ab gene and truncated Bt toxin. PLoS One.7(5) http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0036141 12 Sieradzki Z, Mazur M, Kwiatek K, Swiatkiewicz S, Swiatkiewicz M, Koreleski J, Hanczakowska E, Arczewska-Włosek A, Goldsztejn M.(2013). Assessing the possibility of genetically modified DNA transfer from GM feed to broiler, laying hen, pig and calf tissues. Pol J Vet Sci.16(3):435-41. http://www.ncbi.nlm.nih.gov/pubmed/24195276 13 Swiatkiewicz M., Bednarek D, Markowsi J, Hanczakowska E. and Kwiate, K. (2013). Effect of feeding genetically modified soybean meal and maize to sows on their reproductive traits and haematological indices and offspring performance. Bulletin of Veterinary Institute in Pulaway, 57(3):413-418.

http://www.degruyter.com/view/j/bvip.2013.57.issue-3/bvip-2013-0071/bvip-2013-0071.xml 14 Swindle

MM, Makin A, Herron AJ, Clubb FJ Jr, Frazier KS.(2012) Swine as models in biomedical research and toxicology testing. Vet Pathol. Mar;49(2):344-56 http://vet.sagepub.com/content/49/2/344.short 15 Kararli TT (1995). Comparison of the gastrointestinal anatomy, physiology, and biochemistry of humans and commonly used laboratory animals. Biopharm Drug Dispos. Jul;16(5):351-80. http://onlinelibrary.wiley.com/doi/10.1002/bdd.2510160502/pdf

11

European Society of Toxicologic Pathology Dear colleagues, On the behalf of the European Society of Toxicologic Pathology, I would like to make 2 comments following the GRACE Stakeholder Consultation.

1. My main comment is that much more attention has to be given to the histopathology part of those studies with laboratory animals. I know that you are well aware that biological relevance of the findings is more important than the statistically significant changes. In such studies with laboratory animals, the interpretation of the toxicologic pathologist remains the central part of the safety assessment. The biological relevance of the findings is mainly given by the pathology report together with the clinical pathology conclusion. And the relevance of new tools such Omics still needs the phenotypic anchoring with histopathology data.

Up to now, I consider that we got only a piece of information. During the last Workshop in Brussels, only a partial view of the data was given and the major raw data (histopathology data) were not available. I thank you for your proposal to have access to the slides. But it is note the role of the European Society of Toxicologic Pathology to review the slides. However we will be ready to advise you on that matter as we did when we emphasized the serious inadequacies regarding the pathology data presented in the paper by Séralini et al. (2012) http://www.sciencedirect.com/science/article/pii/S0278691512007880

2. The second and important comment is that studies with laboratory animals are difficult tasks. I am sure that as leaders of the GRACE projects you have in mind that the animal model (In House Historical Data) and the expertise in conducting those studies with appropriate equipment and software) are crucial.

Best regards

Frederic Schorsch ESTP Chairman http://www.eurotoxpath.org/index.php

Bayer S.A.S.

12

Richard Goodman, University of Nebraska, USA 17 May, 2014 GRACE Project 19‐20 May 2014 review. Dear GRACE Participants. Thank you for allowing me to see the preliminary slides and the preliminary report for the 90 day rat study with two varieties of MON810. 1. I believe I made this comment with or at the Berlin, Germany meeting of GRACE. I believe the study design should have included a number of known toxic or anti‐nutrient components to help establish biologically significant differences for many of the parameters. That is even more important in the new general non‐targeted “metabonomic” studies that have just begun. Looking for “statistically significant differences” in a non‐targeted way, or even for the 90 day or chronic rat feeding study blood chemistry and organ weights should identify a number of random differences, as the current data (mostly) seems to show. Therefore these studies lack the most important group(s), positive controls to demonstrate biologically important variation. Since different toxins and anti‐nutrients have different modes of action, one control is not enough. If these studies are to be useful, there must be a biologically relevant basis for setting standards of acceptability. If as in these studies, you are making a large number of measurements from studies with feed that we are pretty darn sure do not have anything wrong in terms of safety or nutrition, what do you think you are measuring? Maybe you are only getting close to measuring the inherent variation in the maize, the diets and the rats. That is NOT a safety issue. 2. The composition of diets and the growth patterns of individual animals will naturally vary some between crop lots and animals. You have a very restricted number of samples. There is some variation, for instance in total protein and even more so in fructose and glucose and total carbohydrates for some of the samples (Pioneer GM and near isoline). One might expect there to be some difference in pancreas weights. There was also a high bacterial count in the GM and near isoline of the A study grain and feed. Wouldn’t you expect that to have some impact on some immune measures? Was the LPS content measured? 3. The data on feed consumption during the last 2 weeks (12 and 13), and especially the last week (13) show an unexpected drop in consumption essentially across the board. What is the case? Is it that the final week was a short‐week? Maybe 6 days instead of 7? Was it because the feed was becoming rancid? Was it because some key component, like a vitamin was declining? Or was it due to a shortage of water? Clearly it also did have a minor impact on body weight gain as well. Were there measurements of the feed at the end of the study to determine if there was spoilage (yes, I realize the diets were irradiated), it seems that rancidity or nutrient decomposition is a reasonable thing to question. Otherwise, water supply. 4. A number of the presentation slides need a brief legend. It is not possible to understand the slides as is. Maybe the speakers will state the measurements, or not. 5. There are no error bars, so it is not possible to see the variation (e.g. protein and sugars in the diets. 6. Chronic study. I believe the OECD calls for more animals per group. You have 10 per group. That was a major criticism of the Seralini study and if memory serves, the numbers should be somewhere closer to 20, although it depends in part on the outcomes you are trying to measure. The Wistar Han is probably a better strain of rats to use compared to Sprague Dawly for chronic studies.

13

7. Antibody measurements against the GM protein. Have you included a design to test antibody production to other dietary proteins? The Cry1A protein is expressed at a very low level in maize. The probability of generating antibodies against it is extremely remote. But you also do not have a standard to measure it against. You should be measuring antibody responses to a number of dietary proteins of different concentrations and characteristics to understand more about the generation of antibodies to dietary proteins. Most humans have some antibodies of different isotypes against some dietary proteins. Although I think it will generally be to more highly abundant proteins. So, is this a waste of time and money since you don’t have a basic understanding of antibody responses to other dietary proteins? 8. More on metabonomics. Does anyone have a dataset that has recorded the variation in rats fed different diets? What happens in the “untargeted” screen if diets vary substantially in specific carbohydrates, fatty acids, total protein, digestible fiber, specific vitamins? I really do not think you can possibly learn much from this study, except some very preliminary data on natural variation under a very restricted set of variables. It seems like a missed opportunity. A lot of money is being spent and animals sacrificed and yet so much more important information might have been gained by altering the study plan to include a few intentionally different diets. Thank you for the opportunity to comment on the Grace project. I am sorry that I am so critical, but I really think that some opportunities to understand normal variation and to set some standards for really harmful food stuffs were missed in this study design. Regards,

14

European Food Safety Authority (EFSA) Questions/comments on 90-day study conducted on MON810 • In the analysis of the test material, it is referred that slight admixture of GM maize was noted in some maize used in the study; can you please comment on this, in particular clarify if the source of contamination has been identified, what are the implications for this study, if any, and which other considerations can be done on this issue in general? •

The sample size used in the experiment, that is the number of experimental units used per sex and per group,was calculated on the basis of SES; in particular SES=1 or less was judged to be of no/little toxicological significance by the Consortium. The number of experimental units in the study was then indicated to be 17 (rounded to 16) assuming a power of 0.8, a significance level of 0.05 and a two-tailed test. However, considering that rats were housed two per cage the number of experimental units should correspond to 8. Clarification is required on definition of experimental unit and on the actual number of experimental units/sex/group in both substudies A and B.

•

More in general, it is considered useful to provide feedback on the appropriateness of SES=1 on the basis of the experience and data from this study (a posteriori analysis) and to possibly extrapolate further to support this approach in 90-day rodent studies inGM applications in the absence of a specific hypothesis.

•

Animal selection for haematology/necropsy workflow: it is not clear from the scheme presented if the sampling for haematology/necropsy was performed per group (i.e whole group 1, followed by whole group 2 etc.) or per cage (cage 1 per each group, followed by cage 2 per each group etc). Please clarify.

•

Please provide information for clinical chemistry sampling workflow.

•

The incidence of histopathological finding presented is very low for a 90-day study in Wistar rat. It would be useful to have recent historical data illustrating background pathology from 90-day studies from the facility. It is recommended to involve the study pathologist and/or the peer reviewer in histopathology of the new 90-day study and in the 1-year study setting and evaluation.

•

The dataset from the two sub-studies indicates inconsistent results in clinical chemistry and organ weights. This should be extensively addressed in the study report.

Questions/comments on the 1-year study on MON810 • Fumonisin contamination is reported for the test material used in the 1-year study. Clarification is needed on the actual material that showed contamination and the potential impact on the ongoing study.

15

Hungarian Ministry of Agriculture

16

17

18

Testbiotech Comment - 18-6-2014 Testbiotech comments on the GRACE stakeholder consultation 19 and 20 May 2014

1. General remarks There are some questions about the experts involved in this part of the GRACE project that need clarification: Apparently, Pablo Steinberg from the University of Veterinary Medicine Hannover is now a core member of the GRACE project team. However, there is no mention of Hannover University as a GRACE partner. Further, the Bundesamt für Verbraucherschutz und Lebensmittelsicherheit (Federal Office for Consumer Protection and Food Safety) is now a GRACE partner. The role of this agency in GRACE remains unclear. How will conflicts of interest be addressed and communicated in the case of Kerstin Schmidt, whose company, BioMath has extensively collaborated with Monsanto? For example, BioMath has for many years been involved in the preparation of Monsanto's post market monitoring reports on the cultivation of maize MON810 in the EU (see for example: http://ec.europa.eu/food/plant/gmo/reports_studies/docs/report_2012_mon_810/report_2012_mon_ 810_farmer_questionnaire_survey_en.pdf).http://ec.europa.eu/food/plant/gmo/reports_studies/docs/re port_2012_mon_810/report_2012_mon_ 810_farmer_questionnaire_survey_en.pdf).

2. Methods Statistical Analysis Statistical analysis of hematological and clinical chemistry data from OECD TG 408 studies on a “cage basis” instead of processing individual animal data is highly unusual, even though it may be in line with EFSA guidance on conducting rodent studies on whole food/feed. Taking into consideration the large amount of statistically significant differences of 11% and 33% GM groups, a re-analysis of numerical individual data should be performed. Feed preparation Was feed analysis performed under GLP? Treatment of maize and Bt content The draft for publication says the maize was milled in Italy, but in his presentation Mr. Kleter says it was milled in Wageningen and there is some mention of pellets in this context. Clarification is needed regarding this process. What was fed to the animals? Pellets or milled maize? Which temperature was used at which level of processing? Please also describe the process of radiation that was used. Which protocol was used in the analysis of Bt content? How was this method evaluated to make it comparable to other publications/ data? At what stage of the experiment did you determine the Bt content?

19

3. Documentation of the results Comparison of plant composition There are a number of differences in the composition of the two MON810 varieties and the isogenic varieties (trypsin inhibitor, protein, carotenoids. ..). These differences should be translated to the EFSA equivalency categories to find out whether one could assume equivalence. Pathology Assuming that TOPALAB submitted a detailed pathology report, this report should be made available

Quality of histopathological evaluation OECD Test Guideline 408 requires the histopathological evaluation of all macroscopic changes seen during necropsy, independent of the dose group of affected individuals. This was obviously not done or at least it has not been reported. This omission becomes obvious when the gross necropsy findings (Table 10) are compared with histopathological findings (Table 11). For instance, inflammatory reaction is mentioned in gross pathology findings (animal Nos. 53 and 57, conventional group 1), but no histopathological correlate has been described.

4. Discussion of the results According to the presentation by Jana Tulinska, there were many significant differences in haematology and biochemistry which were partly sex specific. First, it should be clarified that sex differences must be considered very carefully as it has been shown that these are often underestimated in current research (see for example Clayton & Collins, 2014). Further, careful consideration should be given to the many significant differences in haematological and biochemical parameters. Most differences are not dose-dependent and only appear in one sex. However, it is quite obvious from the slides presented that there is an apparent accumulation of significant effects in the GM groups. In study A, significant differences were found in biochemical parameters in the 11 and 33% GM group each. No significant differences were found in the control. In haematology, significant differences were found in two parameters in the 33% GM group and in one parameter in the 11% GM group. No significant difference was found in the control. In study B, there are eight significant findings in biochemical parameters in the 33% GM group, six significant findings in the 11% GM group. There were no significant differences in the control. In haematology, there were significant differences in ten parameters and eight significant differences in the 11% GM group. No significant differences were found in the control. A slight, but significant decrease in the serum total protein concentration (TP) can be observed in a dose-dependent manner in both studies, although in one sex only in each study, i.e. in males of study A and in females of study B. An assessment of individual animal data revealed five animals with serum total protein values below 50 g/L. Four of them (animal Nos. 1, 6, 283 and 286) were in high dose groups (33% MON810) and one (animal No. 304) in the PR33D48 low dose (11% MON810). Normal serum total protein values are characterized by relatively low variability (coefficient of variation 5-8%). Normal serum total protein values for Wistar rats are in the range between 50 and 70 g/L (e.g. Boehm et al. 2007, Wang et al. 2010, Yamatoya et al. 2012). A decrease in serum protein can have various reasons, including chronic inflammation as related to carcinogenesis (cf. Ohki et al. 2012), and nephrotic syndrome (cf. Palanisamy et al. 2008). To further elucidate possible causes of the

20

hypoproteinaemia observed in animal Nos. 1, 6, 283, 286 and 304, the histological slides of these animals should be re-assessed by a qualified pathologist with emphasis on inflammatory changes, in particular because questions are raised above with regard to the quality of the histopathological assessment.

5. Comments on the one year study Given the contamination of test material with fumosin, the one year study should be canceled and maybe restarted, as it is apparent that the study will not yield results that could clarify the findings of the 90 day studies. In case the study is continued, please elaborate on the necessary modifications in hypothesis, analysis and comparison with the findings of the two 90 day studies.

References Boehm, O; Zur, B.; Koch, A.; Tran, N.; Freyenhagen R.; Hartmann, M.; Zacharowski, K. (2007): Clinical chemistry reference database for Wistar rats and C57/BL6 mice. Biological Chemistry 388: 547–554. Clayton, J.A., Collins, F.S. (2014) Policy: NIH to balance sex in cell and animal studies. Nature, 509, 282 283. http://www.nature.com/news/policy-nih-to-balance-sex-in-cell-and-animal-studies-1.15195. Palanisamy, N.; Viswanathan, P.; Anuradha, C.V. (2008) Effect of Genistein, a Soy Isoflavone, on Whole Body Insulin Sensitivity and Renal Damage Induced by a High-Fructose Diet. Renal Failure 30: 645-654. Ohki S.; Shibata, M.; Gonda, K.; Machida, T.; Shimura, T.; Nakamura, I.; Ohtake, T.; Koyama, Y.; Suzuki,S.; Ohto, H.; Takenoshita, S. (2012): Circulating myeloid-derived suppressor cells are increased and correlate to immune suppression, inflammation and hypoproteinemia in patients with cancer. Oncology Reports 28: 453 458. Wang, X.; Zhang, W.; Wang, Y.; Peng, D., Ihsan, A.; Huang,X.; Huang, L.; Liu, Z.; Dai, M.; Zhou, W.; Yuan, Z.-H. (2010): Acute and sub-chronic oral toxicological evaluations of quinocetone in Wistar rats. Regulatory Toxicology and Pharmacology 58:421–427. Yamatoya, H.; Kawaguchi, H.; Yajima, K.; Kadokura, H.; Yoshikawa, T.; Yamashita, R.; Shiraishi, M.; Miyamoto, A.; Miyoshi, N. (2012): Data on Wistar Hannover Rats from a General Toxicity Study. Experimental Animals 61: 467–476.

21

Appendix 5: Organisations/Individuals Provided With Raw Data On Request

BASF, Germany BfN – Bundesamt für Naturschutz, Germany Eurogroup for Animals, Belgium European Society of Toxicologic Pathology, Belgium High Council of Biotechnology, France Ministry of Rural Development, Hungary Monsanto, USA National Food Agency, Sweden Ramazzini, Italy Rick Goodman, University of Nebraska, USA TestBiotech, Germany VIB, Belgium