Skip to main content
European Commission logo print header

GMO Risk Assessment and Communication of Evidence

Final Report Summary - GRACE (GMO Risk Assessment and Communication of Evidence)

Executive Summary:
The need to reconsider the value of rat feeding trials for the safety assessment of genetically modified (GM) plants (the Implementing Regulation 503/2013 requests for the mandatory performance of 90-day feeding trials whereas this requirement should be reviewed on the basis of new scientific information by the mid of 2016) and the controversial debate about scientific findings on the safety of GM plants were the main drivers for the EU-funded research project GRACE (GMO Risk Assessment and Communication of Evidence). The Project started in June 2012, had a duration of 3,5 years and involved a total of 19 partners from 13 countries.
GRACE picked up the stake by two main work streams: (i) Improving the interpretation of 90-day rat feeding trials, clarifying their added value and exploring alternative approaches to reduce or substitute animal trials, (ii) developing and testing an approach to systematically gather and evaluate the quality of existing evidence on the impact of GM plants. GRACE developed and implemented high standards for stakeholder engagement and transparency in all key stages of the project: during planning as well as interpreting of the results. All GRACE data are available for public scrutiny and debate.

Feeding trials:
GRACE scientists did not find any indication that a routine performance of 90-day feeding studies with whole food/feed would provide additional information on the safety of GM maize MON810 when compared to the compositional comparison of the GM variety and its closest conventional comparator in terms of an initial safety assessment. Moreover, the 90-day feeding trials performed did not reveal any scientific trigger for an extension of the feeding period. Data gathered in the course of an additional 1-year feeding trial are in accordance with the conclusions made in the 90-day trials, i.e. that the feeding of maize MON810 to rats did not lead to adverse effects. Consequently, the 1-year study conducted in GRACE did not provide relevant additional information when compared to the 90-day studies.The data generated by GRACE showed that non-targeted feeding studies may lead to randomly generated significant differences between animals fed with the GM test material and animals fed with a control diet. Such results are not informative for risk assessment. GRACE data support the scientific reasoning that only in case a trigger is available from the initial molecular, compositional, phenotypic and/or agronomic analyses, feeding trials with whole food/feed may provide an added scientific value for the risk assessment of GM crops. Thus, feeding trials might be considered, provided that the study design can be tailored to the posed safety concern. Due to these limitations of rat feeding trials with whole GM food/feed, a mandatory performance in the course of GMO risk assessment cannot be justified in the light of the European goal to replace animal tests and to reduce the suffering of animals.
With respect to alternative studies it was concluded that available methods are very promising but cannot yet replace animal feeding trials in routine GMO risk assessment. In the future, they might help to decide whether an animal feeding trial is necessary and what should be the focus of this feeding study.

Evidence synthesis:
The project also established more comprehensive methods for systematically collecting and assessing existing scientific evidence on environmental, health and socio-economic effects (risks and benefits) of GM plants: systematic reviews and evidence maps. These methods make it possible to draw more robust conclusions on these effects, systematically reflecting the entire body of research available.
GRACE reviews confirmed the conclusions of previous risk assessments of Bt and HT crops with regard to field impacts on the evaluated groups of organisms.

Project Context and Objectives:
There are two major research areas specified for the project:
(a) Reconsideration of the design, execution and interpretation of results of animal feeding trials as well as in vitro and alternative studies for assessing the safety of GM food and feed;
(b) Elaboration and implementation of systematic, transparent and inclusive reviews of existing evidence of potential health, environmental, and socio-economic impacts (risks and benefits) of GM plants (GMPs) or food and feed derived from GMPs.
In addition, GRACE followed an “open science” strategy promoting open-access to research data and enabling stakeholders to become involved in the planning and interpretation of the research.

In 2012 the project was initiated in the middle of the controversy among different stakeholders, from civil society and academia to regulators, about the usefulness and necessity of rodent feeding trials for the risk assessment of foods and feeds derived from GM crops. Moreover, the Implementing Regulation (EU) No. 503/2013 has a mandatory requirement to perform a 90-day (subchronic toxicity) feeding trial with whole GM food/feed, but: “The requirements regarding animal feeding trials in the context of GMO risk assessments should be reviewed in the light of the outcome of this project expected to be available by the end of 2015 at the latest.”
International standards like the OECD test guidelines for toxicity tests with chemicals and more specifically the EFSA guidance on repeated-dose 90-day oral toxicity study on whole food/feed in rodents formed the starting point for the planning of the research. There is an essential difference between testing chemicals and whole food/feed in such feeding trials: Chemicals can be added up to high doses to the feed or water of the test groups to cause a tangible (toxic) effect while plant material cannot be dosed likewise as nutritional imbalances may occur and the amount of feed the animals can ingest is simply limited. As there are intrinsic hurdles in the test design to achieve a clear result informing the risk assessment, the necessity for, and relevance of, such test procedures for GM food/feed safety assessment has reasonably been questioned. The question is whether alternative approaches, which could sufficiently enable the assessment of the potential impact of whole food/feed on human and animal health, are available. The existing assessment routine is based on a comparative assessment of the compositional analyses of the transgenic plant material and its “safe” near-isogenic counterpart as well as specific feeding studies dosing the purified transgenic protein (which can be added to the feed to considerable higher amounts). There is also a strong intention in EU policy to replace or reduce the number of animal tests for various ethical and animal welfare reasons (triple R approach: refinement, reduction, replacement). Therefore, GRACE also investigated alternative methodological options.

The only transgenic crop that is commercially cultivated in Europe (Spain) and available on the market is the insect resistant Bt maize (MON810 maize), expressing an insecticidal protein to fight Lepidopteran pests. Therefore, the project chose MON810 as the transgenic “model” crop for investigations. It was also intended to use transgenic potato provided by USDA to test the study design and added value for another crop, but the prospective project partner withdrew its engagement in the project.

GRACE performed four 90-day feeding trials (subchronic toxicity) with MON810 maize varieties and an extended feeding study lasting 1 year (chronic toxicity) with rats. Near-isogenic varieties were used as controls and additional conventional maize varieties were fed to some study groups to reveal possible variety-linked variability. While the basic design followed EFSA and OECD recommendations, additional samples from tissues, blood and urine were taken to explore the immunological state, allergenicity and impacts on metabolism (metabolomics). In addition, approaches that do not use animal models were employed. Animal cell and tissue cultures were used to examine whether cytotoxicity and inhibition of cell proliferation is caused by different treatments with or without transgenic plant extracts. In addition the plant material was comparatively analysed using standard targeted methods for known nutritional and toxic compounds or non-nutrients, and also using novel “non-targeted” methods screening for changes in gene expression or metabolism.

The European regulatory framework for the market authorization of GM plants and their products demands a comprehensive risk assessment in line with the concept of a knowledge-based bio-economy that shall ensure safe access to the benefits of modern biotechnology. Biosafety research addressing the potential risks of GMP releases and socio-economic research on potential benefits and costs of GMP cultivation has received considerable funding in the EC and worldwide. However, the outcomes are not consistently and comprehensively considered by stakeholders, general public, or in decision making. This may partly be due to several factors:
- the range of studies is extremely broad and rapidly increasing;
- the use of different approaches and methods, and the different ways to present data and results make it difficult to synthesize the findings;
- the value of studies conducted is unclear;
- the debate focuses on uncertainties disregarding the available evidence;
- research results are not “readily available”, conclusions are not transparent or easily understood and may be even biased.
In disciplines such as medicine or social sciences methods for evidence-based decision making – e.g. systematic reviews – have been established. Systematic reviews represent powerful tools to identify, collect, synthesize, and evaluate primary research data on specific research questions in a highly standardized and reproducible manner. They enable the defensible synthesis of outcomes by increasing precision and minimizing bias whilst ensuring transparency of the methods used. This makes them especially valuable to inform evidence-based risk analysis and decision making in various topics and research disciplines.
GRACE adopted the evidence synthesis methodology with regard to the impact assessment of GM crops. Key topics for reviewing environmental, health or socio-economic impacts were identified involving stakeholder feedback. The set of reviews provided by GRACE were designed to increase the awareness of existing evidence on potential risks and benefits of GM crops and GM food/feed and to be of direct or indirect help in the decision making process on current GM crops.
Evidence synthesis methods, especially systematic reviews, are nevertheless demanding with regard to the necessary resources (workload, time). To promote the methodology while reducing the necessary efforts GRACE developed online tools to facilitate cooperative performance of evidence synthesis in remote review teams.

A primary objective of GRACE was to improve the accessibility and presentation of the scientific information in order to provide better understanding for all stakeholders and the general public. Therefore, a central, open-access web portal called CADIMA (Central Access Database for IMpact Assessment of crop genetic improvement technologies) was established to
• facilitate the exchange of general information about the impact assessment of GM organisms by providing links to international databases / data providers;
• support the performance of evidence syntheses, the exchange and update of reviews;
• collect data from feeding trials performed by different projects to support joint analyses.
CADIMA will permanently operate after the project period.
Furthermore, the project aimed to implement an active stakeholder involvement in the planning and the interpretation phase of the project. The feedback from stakeholders allowed the project to address their requirements and improve the study/research designs as far as possible. Moreover, stakeholder requests that could not be realized (e.g. due to limited resources) as well as controversial discussions were documented to allow future consideration of the issues raised.
The main objectives of the communication strategy of the project were:
- to provide open-access publications of the project results;
- to interact with complementary projects;
- to broadly report on the project and follow-up activities.

Project Results:
The project was structured in 10 work packages (WP). Two work packages investigated the design and added value of rodent feeding studies with whole GM food/feed (WP 1) and alternative methods and approaches (WP 2). Three work packages conducted reviews exploring impacts of GM crops and derived products on human health (WP 3), socio-economic factors (WP 4) and the environment (WP 5). WP 8 explored methods for evidence synthesis to inform GMP risk assessment. WP 6 established an internet portal to provide a central access for information sharing. Stakeholder involvement was organised by WP 7, while the general communication and dissemination was the task of WP 9. WP 10 was responsible for the project management.

Feeding trials and alternative approaches
Fulfilling one key objective of the project WP 1 and 2 focussed on the reconsideration of the design, execution and interpretation of results of animal feeding trials as well as in vitro and alternative studies for assessing the safety of GM food and feed.
Transgenic and non-transgenic maize varieties were grown in Spain in 2012 and 2013. The maize (kernels) was purchased from the growers. The project had not signed any obligations concerning the use of the event or the varieties.
The following studies were performed:
• Study A: 90-day study with Monsanto MON810 maize (5 male + 5 female groups)
• Study B: 90-day study with Pioneer MON810 maize (5+5 groups)
• Study C: 1-year study with Monsanto MON810 maize (4+4 groups)
• Study D: Longitudinal and metabolomics study with Monsanto MON810 maize (3+3 groups)
• Study E: Longitudinal and metabolomics study with Pioneer MON810 maize (3+3 groups)
For study A and B the following experimental steps were undertaken: Within the trial period the body weight and the feed consumption were monitored, clinical and ophthalmological observations were recorded. At the end of the study haematology and clinical biochemistry parameters were quantified, a gross necropsy including the determination of the absolute organ weights as well as a histopathological analysis were performed. Tissue samples were taken for the analyses of protein, metabolite and gene expression patterns (proteomics, metabolomics, transcriptomics) and plasma samples were taken for further metabolomic analyses. In contrast to studies A and B the studies D and E were performed as pilot studies (less animals per group) with additional blood and urine sampling to follow the development at certain time intervals and to record immunological variables and metabolic dynamics (metabolomics).

For study C the following experimental steps were undertaken: Within the trial period the body weight and the feed consumption were monitored, clinical and ophthalmological observations were recorded. Haematology, clinical biochemistry and urinalysis parameters were quantified after 3, 6 and 12 months. At the end of the study a gross necropsy including the determination of the absolute organ weights as well as a histopathological analysis were performed. Tissue samples were taken for the potential analyses of protein, metabolite and gene expression patterns (proteomics, metabolomics, transcriptomics) and plasma samples were taken for potential metabolomic analyses.

Balanced diets with an in toto maize inclusion rate of 33% were fed to the test animals. Two dose levels were tested for the transgenic maize variety (inclusion rate of 11% and 33%) The composition of the feed was analyzed with regards to nutritional value, non-nutrients and toxins.
The design of the 90-day feeding studies was based on the OECD test guideline for rodent toxicity studies No. 408 and the recommendations of the European Food Safety Authority; the design of the 1-year feeding study was performed with regards to the OECD test guideline No 452.

The statistical analyses were based on the comparison of the test groups with the control group fed with the near-isogenic variety. The project followed the suggestion of EFSA to analyse “Standardized Effect Sizes” (SES) and compared it with the usual statistical procedures (ANOVA). Based on the experiences gathered by GRACE, the SES approach eases the interpretation and contextualization of the results by providing an overall picture of potential group differences and the associated magnitude of difference. Predetermined deviations of toxicological concern can be co-displayed within the SES graphs (Schmidt et al. 2015, Archives of Toxicology; DOI: 10.1007/s00204-015-1487-8).

Results of feeding trials with whole food/feed
Though a considerable number of statistically significant differences for blood biochemistry and haematology variables and the weight of some organs between the test groups fed transgenic maize and the corresponding control group in trials A and B were observed, the differences were interpreted as being unrelated to the MON810 event. These results could neither be reproduced in studies D and E nor within the 90-day period of the 1-year trial. This was also confirmed by analysing the consistency and reproducibility of statistically significant effects in the 90-day data of all five studies. In addition the extended test period of 1-year did not reveal a toxicologically relevant pattern related to the test groups. Gross necropsy and histopathological findings were rare and did not show any group specific effects. When interpreting the results from animal feeding trials conducted in the course of GRACE, all statistically significant differences between groups were evaluated for their toxicological relevance. The varying patterns of significance in the different studies, as well as the discussions arising from the publication of the results from Studies A and B (Zeljenkova et al. 2014, Archives of Toxicology 88; 2289-2314.), highlighted the importance of the distinction between the statistically significant difference of test groups and the toxicological relevance of such differences, i.e. a single statistically significant difference is not necessarily indicative of organ or tissue specific toxicity, which generally becomes evident through changes in several related parameters.
Studies A and B did not reveal any scientific trigger for an extension of the feeding period.
The data gathered in the course of Study C (1 year) concur with the conclusions made after 90 days that administration of maize MON810 to rats did not produce adverse effects. In addition the extended study conducted in GRACE did not provide any relevant additional information compared with the 90-day studies.
The GRACE data did not provide any indication that the performance of 90-day feeding studies (following OECD or EFSA guidelines and current practice) with whole food/feed would provide additional information on the safety of maize MON810 when compared to the compositional comparison of the GM variety and its closest conventional comparator in terms of an initial comparative safety assessment. These findings are in line with the EFSA Guidance Document that forms the basis of the current Implementing Regulation 503/2013.The data also showed that such non-targeted feeding studies may generate outcomes at the level of the variability of the lab, i.e. generating significant differences randomly. Such results do not inform the risk assessment nor do they increase the confidence in the data provided.
When a trigger value is recorded (e.g. from the initial molecular, compositional, phenotypic and/or agronomic analyses) a targeted hypothesis can be developed to tailor the study design to the posed safety concern. In such a case feeding trials with whole food/feed may provide added scientific value for the risk assessment of GM crops (EFSA, 2011).

Results of omics analysis of animal tissues and fluids
The inclusion of additional parameters (added to those stipulated in the OECD guidelines) in the trials performed throughout GRACE showed that immunotoxicity testing, allergenicity assessment and “novel” ~omics techniques principally enable a detailed examination of a broader spectrum of additional parameters, but the approaches are currently not validated or standardised. Moreover, planning and interpreting the studies should consider thoroughly the biological variations within the tested parameters.
In the GRACE experiments, targeted serum screening was applied for the assessment of allergenicity and immunogenicity of GM vs. near-isogenic control maize and commercial varieties. Immunosuppression was examined by phenotypic analysis of leukocytes in rat tissues, and three in vitro immune function assays. There was high inter-individual variability in the animal immune responses, as expected, which might limit the applicability of these assays.
In case of targeted approaches for the examination of animal tissues, the integration of omics screening of potential target organs provides a basis for the identification of causal molecular patterns associated with toxicity. The identification of pathways of toxicity may unravel mechanistic details that enhance the regulator’s confidence in classical parameters by providing a mechanistic basis for decision-making. Intestinal tissues from rats, which had primarily been exposed to the ingested feed materials, were analyzed. These tissues have important functions in terms of immunology, biotransformation and signalling responses to xenobiotics. Bioinformatics analysis of the transcriptomics data (from studies A and B; > 25.000 genes) revealed a clear temporal pattern in expression of genes being related to the sampling time. Nevertheless perturbations of pathways of toxicity have not been detected, underlining the study results revealed by classical OECD endpoints.
The longitudinal study design (studies D and E) has the potential to detect progressive changes when compared to single endpoint measurements: no such changes in the longitudinal metabolomics study in the GRACE project were detected; again this reinforces the finding that no toxic effect could be detected during the sub-chronic studies.
The experimental data underline that an application of transcriptomics/proteomics approaches without a targeted and testable hypothesis as well as adequate controls did neither facilitate hypothesis generation nor did it reveal unintended effects of feeding MON810 maize. Unintended effects were also not detected in metabolomic analyses with sera from the 90-day studies.
Taken together, GRACE data indicated that omics approaches detect the coordinated homeostasis of the examined tissue(s) which in turn may facilitate the detection of nutritional or toxicological perturbations. Variations in experimentally controlled and in non-controlled factors were trackable (e.g. time of sampling, linked with circadian rhythms, heterogeneity of sampling related to the cellular composition or anatomic sub-sites, etc.). Therefore the experimental design should be based on clear-cut questions and optimized for omics approaches and a careful evaluation of the available data must be ensured to prevent misinterpretations. A targeted and testable hypothesis would be necessary to inform the decision about the choice of specific and associated parameters to be included in the trial.

Results of cell-based assays
In the course of GRACE in vitro test systems with primary and permanent intestinal cell cultures (pig, rat) were explored. An essential step was to optimize plant extracts for this application. Aqueous extracts were prepared using basal media of the respective cell cultures, and dose tests were performed for optimization. Quantification of Cry1Ab protein in maize extracts revealed that exposure levels achieved in vitro can be manifold higher than published concentrations in in vivo digesta. The validation of novel in vitro tests requires the comparison to available in vivo data. Therefore, targeted feeding trials with whole food/feed with toxicological potential are needed to prove the validity of the developed test. Currently, there are no in vitro methods for complex endpoints like repeated dose systemic toxicity and the development of such assays will require major scientific advances.

Results of omics on plant material
GRACE examined omics methods for the comparative assessment of transgenic and non-transgenic plants which could supplement detailed chemical analyses of plant and feed material for its nutritional value, non-nutrients and toxins. However standardized and validated test procedures for using omics are currently not available.
A “one-class model” approach was used: the GM variety is compared to its closest conventional comparator, as well as to a range of conventional varieties by impartial statistical means. It was shown that the one-class model classifies “positive controls” such as a mycotoxin-contaminated maize sample (additionally sampled in Spain, not related to the maize materials that have been fed to the test animals) as outside of the ‘safe maize’ class. Similarly the “one-class model” classified experimental potato varieties that are fit for human consumption but genetically more distant from the lines that are currently consumed as outside of the ‘safe potato’ class. All maize varieties used in the feeding trials A, B, D, and E were classified as inside of the ‘safe maize’ class. Due to time limitations, the maize harvest included in the diet from study C was not analyzed.
Furthermore, GRACE omics data have shown that elements with a foreign origin such as the Cry1Ab gene and its expressed product in the MON810 maize can readily be identified by following this non-targeted approach. Based on these observations, it can be stated that the omics data generated provide qualitatively structured details of the plant material which facilitates a non-targeted “safety“ evaluation. Thereby it provides a better basis for the decision on the scientific rationale to frame the subsequent risk assessment steps, which may include the performance of an animal feeding trial with the plant-derived whole food/feed.

Overall conclusions
Based on a scientifically justified design, a vast amount of animals had to be sacrificed in the course of GRACE [Studies A and B (90 days): 160 animals/study, Studies D and E (90 days): 120 animals/study, Study C (1 year): 160 animals, in total: 720 animals]. Considering the available scientific evidence and the intrinsic limitations of a feeding trial with whole food/feed, the GRACE team concluded that mandatory performance cannot be justified in the light of the RRR approach (replacement, reduction, and refinement of animal tests). GRACE explored and suggested complementary and alternative approaches that may be further elaborated to inform the development of a targeted hypothesis for assessing whole food/feed and to support a targeted study design to address the specific safety concern more appropriately. The expected magnitude of a distinctly identified potential effect should be included into the test hypothesis and guide the decision on whether a feeding study should/could be performed to achieve a valuable information gain.

Synthesising evidence from GMO impact data
The second key objective of GRACE was the elaboration of systematic, transparent and inclusive reviews of existing evidence of potential impacts of GM plants or food and feed derived from GMPs on health, environment, and socio-economic factors. In the course of GRACE, 7 systematic reviews and 7 evidence maps were conducted for prioritized research questions in order to summarize the available evidence base related to potential impacts caused by the deliberate release of GM crops on human/animal health, the environment and socio-economy. The specific questions being addressed and the evidence synthesis approach applied are as followed (SR=systematic review, EM=evidence map):

Environmental impacts
• Does the growing of Bt maize change abundance or ecological function of non-target animals compared to the growing of non-GM maize? SR
• Are population abundances and biomasses of soil invertebrates changed by Bt crops compared with conventional crops? SR
• Are soil microbial endpoints changed by Bt crops compared with conventional crops? SR
• What is the evidence on the inheritance of resistance alleles in populations of lepidopteran/coleopteran maize pest species? EM
• How susceptible are different lepidopteran/coleopteran maize pests to Bt-proteins? SR
• What are the effects of the cultivation of GM herbicide tolerant crops on botanical diversity? SR
Impacts on human/animal health
• What are the characteristics of studies on the risk of allergic sensitization and elicitation in humans and animals exposed to an allergenic crop plant that has been genetically modified as compared to individuals exposed to its non-GM crop plant counterpart? EM
• What are the characteristics of comparative studies of changes in the assessment of levels of key chemical crop constituents in GM crops compared to non-GM crops? EM
• What are the characteristics of toxicity studies in which animals receiving newly expressed proteins from GM crops are compared to animals that are administered appropriate controls without these proteins? EM
• What are the characteristics of repeated-dose feeding studies in which experimental animals receiving whole food or feed products derived from GM crops are compared to animals receiving conventional non-GM counterparts? EM
Socio-economic impacts
• What are the socio-economic impacts of GM crops worldwide? EM
• What is the impact of the introduction of GM crops on the welfare effects in different countries in comparison to a situation where there are restrictions on GM cultivation? SR
• What is the impact on GM regulation of different political actors and other drivers in the EU in comparison to the situation in the US? EM
• What is the impact of trade restrictions of GM products in different countries on the competitiveness of different partner countries and corresponding sectors in comparison to a situation where there are no restrictions on GM trade? SR

Furthermore, the adequacy of systematic reviews and evidence maps to inform GMO impact assessment and to facilitate the communication of impact data were assessed. The major results related to aforementioned activities are laid down in the following section. It has to be emphasized that, due to the fact that a cost neutral prolongation was not granted by the EC, only preliminary results can be provided at this stage for the conducted reviews as the final manuscripts are currently under elaboration.

Evidence synthesis on potential environmental impacts (WP5)
Bt Maize impacts on arthropods (systematic review)
Preliminary results show that
• Untreated Bt maize had similar invertebrate communities than untreated non-Bt maize. Only few significant differences were observed and those were usually traceable to individual studies.
• Treatments with broad spectrum insecticides caused several significant effects on invertebrate populations, in particular foliar applications of pyrethroids.
• Results mainly confirm results of earlier meta-analyses in this area.

Bt Crops impacts on soil invertebrates (systematic review)
Preliminary results show that
• No significant effects of Bt crops on soil invertebrates were observed.
• No effects of individually selected Cry protein types could be detected.
• No particular taxonomic orders appear to be affected by Cry proteins.
• The estimated time of maximum effect of Cry1 was quite variable, which was to be expected according to the estimated small and variable effects.

Bt Crops impacts on soil microorganisms (systematic review)
Preliminary results show that
• Bt Crops do not have major effects on soil microbial endpoints as compared to conventional crops.
• Minor short term effects on the microbial communities might exist, however they hardly have any functional significance, as compared to other factors as plant species and growth stage, soil type and time.

GMHT Crops impacts of management on weeds (systematic review)
Preliminary results show that
• Effects on weed populations depended on crop, contrasted herbicide programs, exposed weed species and types, and a range of other management factors e.g timing, frequency, rates, dose.
• GMHT treatments were more effective on grass weeds in maize and broadleaved weeds in soybean and cotton.
• Generally glyphosate treatments reduced weed populations in more situations/comparisons, but there were always some comparisons where conventional herbicide programs produced similar effects to glyphosate.
• Glufosinate treatments tended to produce more variable responses in weed populations, especially if applied alone.

Inheritance of resistance (evidence map)
Preliminary results show that
• Most information was available for Helicoverpa armigera, followed by Ostrinia nubilalis.
• For all other species, only single data sets were found.
• The data set is very limited due to the few cases of resistance evolution for the relevant species/crop combinations and the IR ranged from recessive to non-recessive depending on pest/crop system.

Baseline Susceptibility (systematic review)
Preliminary results show that
• Most data are available on Helicoverpa armigera and Cry1Ac and have limited relevance to Bt maize in Europe.
• Variability of data is high due to differences in toxin resources, experimental set ups and measured end points.
• Data availability is unequally distributed between species.
• The acquired data set gives a good overview on available data and may be an important basis to model and assess the potential for evolution of resistance of different crop-Bt-protein-species combinations.

Overall conclusions
• GRACE reviews confirm the conclusions of previous risk assessments of Bt and HT crops with regard to field impacts on the evaluated groups of organisms.
• They provide complementary scientific information that may inform risk assessors and managers, and those involved in environmental monitoring and integrated pest management.
• They provide weight of evidence information that may inform policy and decision makers.
• Data included in applications could not be used due to intellectual property rights.

Evidence synthesis on potential impacts on human/animal health (WP3)

Comparative analysis of key chemical constituents in GM crops (evidence map)
Preliminary results show that
• A great variety of crop species and GM events were identified.
• Various categories of targeted analytes were covered.
• Identified omics studies mainly focused on the analysis of metabolites.

Assessment of the endogenous allergenicity of GM crops (evidence map)
Preliminary results show that
• Studies were mainly performed on mouse models and allergic patients.
• Two general studies types were identified:
o Studies assessing GM crops for changes in endogenous allergens as potential unintended effects;
o Studies reporting the application of RNAi technology to lower the level of one or more endogenous allergens in the GM crop as an intended effect.

Toxicity studies with newly expressed proteins from GM crops (evidence map)
Preliminary results show that
• A diverse range of proteins were tested with a preponderance of Cry proteins and insect resistance traits in general.
• Rodents appear to be the main experimental animal while livestock and zebrafish were occasionally used.
• Classical toxicological endpoints were mainly detected.

Feeding studies with whole food/feed derived from GM crops (evidence map)
Preliminary results show that
• Most studies were on GM corn.
• Rat and poultry appear to be the main experimental animals.
• Most frequent study duration ranked between 30-60 and 85-95 days.

Overall conclusions
• The drafted evidence maps on the four health-related topics (toxicity of newly expressed proteins and whole foods/feeds, allergenicity, composition) showed that publications cover a wide range of crops, newly introduced traits, experimental animal species and other experimental models and parameters.
• Evidence maps can inform both risk assessment and risk management communities.
• The references retrieved through the search actions but de-selected during further selection and extraction, may also provide interesting insights, such as the observation that a high number of studies have been published on the production and characteristics of oral vaccines and other pharmaceuticals in GM crops (used as “plant factories”), and the seemingly increasing number of publications by Chinese and other Asian authors in recent years (e.g. in non-English literature).
• Data included in applications could not be used due to intellectual property rights.

Evidence synthesis on potential socio-economic impacts (WP4)

Socio-economic impacts at micro level (evidence map)
Preliminary results show that
• Farm- and consumer-level topics comprised the majority of the studies.
• Asia and North America provided the largest evidence base.
• Cotton and corn were the most studied crops.
Study methodologies were mostly observational and research designs non-experimental.
Sectoral and macro-level impacts of GM crops (systematic review)
Preliminary results show that
• GM crops have impacts on aggregate welfare change. Economic benefits of GM crops are positive and significant.
• On average, producers are estimated to capture the largest share of the change in economic benefits, followed by the biotechnology firms and consumers.
• A number of study characteristics could be identified that help to explain the highly heterogeneous impact across the selected studies.

Trade impacts of GM crops (systematic review)
Preliminary results show that
• Mixed results for trade flows and production were retrieved. Findings on the price component seem to be in line with empirical literature.
• A number of study descriptors which help explain the presence of heterogeneous impact across the selected studies could be identified.
• The impact size seems to vary by geographic region, sectors, type of model used and whether a publication has undergone a peer-reviewed process.
• Most literature tends to focus on the impact at country (and economic region) level, whereas there is an evidence gap on the impact on competitiveness of trade-related measures on different actors of the supply chain.

Politics of GM crops (evidence map)
Preliminary results show that
• No single explanation accounts alone for the difference in GM legislation in the EU and US.
• Some studies have related differences in GM regulation to differences in consumer preferences and citizen’s trust between the EU and the US, triggered by the food crises in Europe in the 1990s and 2000s.
• The media seem to have a pivotal role in influencing food risk perception and views on GM food.
• Other studies have identified, in comparative advantage factors between EU farmers and agrochemical industry and US producers and biotech companies, a key factor in explaining why the regulatory differences may continue to exist in the long run. The importance of heterogeneity of interests within each group of actors has been also emphasized.

Overall conclusions from the reviews
• Evidence gaps and areas for further research have been identified in terms of geographical focus and research fields (e.g. supply chain, environmental economics, food security, distribution impact of trade-related measures among actors).
• GRACE team created an extensive and unique database on socio-economic impacts of GM crops that serves as a solid foundation and guide for future research.
• Preliminary conclusions revealed that the introduction of GM crops has impacts in terms of aggregate welfare change and that there are mixed results on trade-related impacts of GM crops.
• A number of factors were identified explaining heterogeneous impacts across the selected studies. Evidence synthesis approaches are, if feasible, particularly valuable tools when informing decision-making.
• Relevant statistics are often not reported and this might affect the way a quantitative data synthesis is performed. For future research on socio-economic impacts of GM crops it is recommended that more comprehensive statistical information on the data used should be provided.

Can systematic reviews and evidence maps inform GMO impact assessment and facilitate the communication of impact data?
Systematic reviews and evidence maps may provide added value when informing different risk assessment or risk management steps about GMO impacts, by increasing precision, minimizing bias and allowing for active stakeholder involvement. However, due to associated limitations, such as the availability of primary data and the strong demand on resources, a decision about their conduct has to be made on a case-by-case basis and thus a mandatory systematic review as required by the Implementing Regulation (EU) No 503/2013 would not be feasible. Further details are provided by “Can systematic reviews inform GMO risk assessment and risk management? Frontiers in Bioengineering and Biotechnology, 2015, DOI: 10.3389/fbioe.2015.00113”
Based on the experiences of GRACE, systematic reviews and evidence maps can facilitate the communication of GMO impact data by:
o allowing transparent discussion of assessment criteria by means of the review protocol;
o clarifying exactly which evidence has (or has not) been included;
o assuring the traceability of review conclusions.
As systematic reviews evaluate the validity of included studies, they can provide feedback on how to raise the study quality.

Stakeholder involvement (WP7)
The GRACE stakeholder approach is distinct from similar exercises because it provided and supported stakeholder involvement not only in discussing project results but also in the planning of research and other activities (upstream engagement). In pursuing this approach, transparency and accessibility of project information, results and data became key issues because they allowed tracking of what was being done, how it was done, by whom it was done and what results have been obtained. Considerable efforts were taken during the project to make such information available to stakeholders – in particular in the context of stakeholder consultations. Moreover, by systematically providing responses to their written comments, stakeholders could track how their comments were considered – or why they were not considered.
A broad range of stakeholders were contacted for each open consultation step encompassing 500 to 1,300 contacts and each stakeholder interested could participate. The only restriction was applied to journalists in order to maintain a discussion forum where everyone could speak freely. In the course of the project a total of 143 individual stakeholders from 19 EU Member States, the EU level, Switzerland, Norway, the USA and international organizations, participated in one or several consultations. All workshops had a very good attendance – between 25 and 54 stakeholders. In all key consultation steps all main stakeholder groups (competent authorities, industry, civil society organizations, researcher) were represented and actively contributed. In addition GRACE received and responded in writing to ca. 1,100 written comments from stakeholders.
Overall, five large stakeholder workshops and eight written consultations were conducted with activities and results documented in great detail in ten consultation reports all of which are available at the project website.

Beyond the specific issues dealt within GRACE the project also allowed for exploring the possible gains and challenges in implementing extended engagement and increased transparency in GMO risk research in general. Participant’s feed-back was generally positive to very positive on how these aspects were implemented. Discussions evolved over time from an adversarial to a much more constructive style, trust increased. Many stakeholders’ comments were adopted and partly helped to improve the research. On the other hand it became clear that putting a meaningful and extended stakeholder engagement and high level of transparency into practice would benefit from supportive measures. More flexibility would be needed in research projects for modifying research plans and time schedules in order to consider stakeholder inputs. Dedicated fora would be needed for addressing non-scientific / non-technical aspects raised. Procedures and reassurance would be needed for scientists revealing non-published data and results to stakeholders without conflicting e.g. with scientific publisher requirements. Stakeholder might sometimes need support from project resources, as engagement and constructive contributions requires a lot of time and travelling.
These and other more general insights won in GRACE are particularly relevant and topical as they provide (practical) guidance when implementing the core concept dimensions of Responsible Research and Innovation which is not only underlying the European Commission research funding program Horizon 2020 but also many national programs.

Further dissemination activities (WP9)
Following the establishment of the GRACE website as the central dissemination hub, the number of visitors has increased continuously, with an average visit of about 2000-3000 (up to 3500) per month and with up to 6000 page views for the last 6 months. The average visit duration (3 min 30 s) indicates a high visit engagement and interest in the provided information. There are about 940 external backlinks found on referring external pages that link to www.grace-fp7.eu. Accordingly, the Ahrefs Domain Rank is comparatively high (54, on a scale of 1 to 100). In addition a dissemination network (contacts to national stakeholder organisations) was established, covering 1100 contact persons from European consumer organisations, farmer associations, relevant food safety authorities, environmental organisations, and animal welfare organisations. Furthermore, 1500 stakeholder contacts (industry, industry associations, scientists, participants of other EU projects, interested persons registered via the project website) were used to inform the public via press releases, open letters and newsletters. The dissemination of GRACE related information has been enhanced through the use of online-PR portals (Marketwire / PR gateway / PRLOG; distribution platform to approx. 550 other portals; resulted in between 1200 and 1500 new referencings per press release) and direct mailings to the above mentioned dissemination network, other stakeholder contacts and media contacts (about 650 journalists; categorized by zimpel® as active in the area of GMO products, GMO food/feed, agricultural politics, environmental issues). In addition, videos from the presentations given during the final conference are being made available to the broader public via the establishment of a YouTube channel.

The open access database CADIMA (WP6)
In the course of the project, the open-access database CADIMA (Central Access Database for Impact Assessment of crop genetic improvement technologies) was established togrant access to raw data generated in the course of GRACE and to support the conduct of systematic reviews and evidence maps and facilitate the dissemination of evidence synthesis results. CADIMA will be permanently established at and maintained by JKI and thus serves as an ideal tool to support 1) synergistic effects between research projects and 2) a centralized dissemination of project results. Besides GRACE, the research projects PreSto ERA-Net, G-TwYST and GMO90+ agreed to use CADIMA when making research results publically available.

Potential Impact:
The two main research tasks defined for the project GRACE (1. The reconsideration of the design and added value of animal feeding trials and alternative approaches for assessing the safety of GM food and feed and 2. The performance of systematic, transparent and inclusive reviews of existing evidence of potential health, environmental, and socio-economic impacts of GM plants (GMPs) or food and feed derived from GMP) placed GRACE at the centre of debates on the safety assessment of GMPs involving risk assessors, regulators, policy makers, industry, scientists, civil society organizations and the wider public. The heterogeneity of the different stakeholder groups, their specific scopes and interests in the research topics and the defined limitations of the project (i.e. only prioritised research questions could be investigated), led the project team to implement an “open science” strategy enabling open access to research results including data to support further discussions and integrate investigations and decisions. Moreover, stakeholders were involved in the planning and interpretation of the research and their contributions were thoroughly documented for further consideration.

Regulatory issues
The European Commission Implementing Regulation 503/2013 states that “The current uncertainties in relation to the need and design of 90-day feeding trials will be addressed by a large research project under [...] the seventh Framework Programme for Research (FP7). The requirements regarding animal feeding trials in the context of GMO risk assessments should be reviewed in the light of the outcome of this project expected to be available by the end of 2015 at the latest.” This explicitly places the products of the GRACE project as contributing to the development of the regulatory framework. In November 2015 the project introduced its final conclusions and recommendations to the broader public. Prior to that the collated but unpublished data of the feeding trials were made available to, and discussed with interested stakeholders in October 2015. The publications and presentations already initiated a discussion among stakeholders about the general conclusions that non-targeted (90-day) rodent feeding trials with whole GM food/feed do not per se provide an information gain for the risk assessment, but may produce random and hence misleading findings.
The results underline the recommendations and guidelines for the design and performance of animal feeding studies provided by EFSA and the OECD. Nevertheless, the results also show that a thorough design may not overcome intrinsic limitations of non-targeted studies with whole GM food/feed.
Thereby the project revealed a possible contradiction between the mandatory performance of a 90-day feeding study with whole GM food/feed requested by the Implementing Regulation EU 503/2013 and the “Triple R” approach targeting at reducing animal feeding trials with limited value.

Regulatory decision making
In the field of health, environment and socio-economics, evidence maps and systematic reviews provide a transparent overview of the evidence base and the integrated results from primary research data but also provide insight into information and knowledge gaps. The use of these tools to support regulatory decision making on GMOs was investigated in GRACE.
The exploration of the evidence base on GM crop impacts collated and analysed (primary) study data to answer several research questions (systematic reviews) and provided an overview of existing data sources in a broader safety context (evidence maps). The set of reviews conducted in the course of GRACE confirmed the previous assessments on impacts of GM crops authorized in Europe.
The adoption of the methodology of systematic reviews (SR - systematic evidence synthesis) can improve science-based decision making by providing defensible syntheses of outcomes from primary research data and by increasing precision and minimizing bias whilst ensuring transparency of the methods used. The standardised methodology, the comprehensive documentation, and the transparency and stakeholder engagement requirements can provide risk assessors and decision makers with a more robust synthesis of existing evidence and can help to clearly communicate the processes of evaluation and the final conclusions. In addition SR allows review update and re-evaluation of the evidence base once new information becomes available. The SR evidence base can also document data gaps or methodological restrictions. The project also demonstrated when there is a need to perform systematic reviews and when they are not appropriate. A detailed evaluation of benefits and challenges of using SR methods to inform GMO risk assessment and risk management is provided in the publication by Kohl et al. 2015, Frontiers in Biotechnology DOI: 10.3389/fbioe.2015.00113. SR demands specific research questions and sufficient primary research data to allow a quantitative impact analysis. However, the European Commission Implementing Regulation 503/2013 requests a “systematic review” covering all issues relating to the risk assessment of a GMP. The GRACE project showed that the standard methodology of systematic reviews is not applicable to this request by the Implementing Regulation and that further clarification of terminologies and scopes is urgently needed.

Economic dimensions
The project identified regulatory burdens of the Implementing Regulation EU 503/2013: the mandatory 90-day feeding trails with whole GM food/feed and the obligation to perform a systematic review including all potential effects on human and animal health covered by the application. Both obligations increase the costs of market releases of GMP without providing a clear benefit (i.e. an information gain) for the risk assessment. Since these economic consequences have been identified, the requirements regarding animal feeding trials in the context of GMO risk assessments as described in the Implementing Regulation should be reviewed in the light of the outcome of GRACE.
A comprehensive overview on international studies about socio-economic impacts (risks and benefits) of GMP is provided in an evidence map, and further systematic reviews and evidence maps covering welfare effects, trade restrictions and policy impact comparing Europe and countries outside Europe. The material provides starting points for further detailed and focused consideration and research.

Societal and political dimension
The research topics of GRACE refer to societal and political concerns about the risk assessment of GM crops as well as animal welfare. The science-based risk assessment as a fundamental concept in EU regulations relies on the confidence of the society as well as policy and decision makers that the methods are appropriate. GRACE’s findings indicate that the legal provisions discussed above are scientifically questionable. Inappropriate assessment methods undermine the reliability and credibility of the risk assessment. Performing animal feeding trials without science-based reasoning contradicts the EU regulations on animal welfare (Triple R approach). GRACE explored several alternatives and the stakeholder consultations provided detailed discussions on these issues. They are described in a comprehensive document accessible at the project website (www.grace-fp7.eu).

Scientific advancements
Though rodent feeding trials are well established for testing the toxicity of chemicals, they have not been fully validated for assessing the impact of whole GM food/feed. The project revealed several limitations when adopting OECD standards and EFSA recommendations. Beyond the consideration of the basic study design and its added value the project investigated several supplementary methods to record immunological and allergenic effects or to explore “novel” ~omics techniques. These methods may not only complement feeding studies with GMP but may be adopted in a broader context. Nevertheless, further investigations are necessary for validation and standardisation. Similarly, the validation of novel in vitro test systems introduced by the project requires the comparison with available in vivo data in order to be considered as alternatives to animal feeding studies.
The project also suggested improved statistical methods for the analyses of feeding trial data. The traditional ANOVA approach was compared to a “modern” linear mixed-model (LMM) approach for analysing periodically recorded parameter (feed consumption, body weight), and the use of standardized effect sizes was implemented as well (Schmidt et al. 2015, Archives of Toxicology DOI: 10.1007/s00204-015-1487-8).The LMM approach results only in one statement on differences in e.g. the weight development between study groups. Standardized effect sizes are calculated for further endpoints. The parameter values are standardized, allowing the comparison of the group differences in the whole study at a glance. Moreover, standardized effect sizes, statistical significance and the toxicological relevance (toxicologically defined thresholds) can be displayed simultaneously in the same graph. Both methods improve and simplify communication of the statistical findings.
Detailed omics analyses of plant material might be considered as being more informative to screen for unintended effects of plant breeding than animal feeding trials with whole foods. A strategy to analyse data from non-targeted omics characterisation of plant material as an omics-based comparative assessment was tested employing a ‘one-class’ classification tool developed by RIKILT Wageningen UR in collaboration with Wageningen UR Biometris (the statistical group of Wageningen UR) and by the chemometricians of the University of Nijmegen (department of Analytical Chemistry). The one-class classification tool aims to assess omics profiles of new plant varieties in the light of a set of similar profiles that have been obtained from a series of commercially available varieties that are considered as safe. Further tests with known “unsafe” material are needed for detailed elaboration and validation of the approach.

The project also explored the evidence synthesis methodology (systematic reviews and evidence maps) to support the impact assessment of GM crops. Controversies over the quality of peer-reviewed papers on potential impacts of GMPs underline the necessity for transparent and unbiased scientific reviewing. This is essentially facilitated by systematic reviews. GRACE provided a set of systematic reviews and evidence maps successfully adapting the methodology and demonstrating options to integrate stakeholder consultations into the review process in a unique manner. Since systematic reviews are demanding with regards to time and resources GRACE established the internet portal www.cadima.info providing a free web-based tool to support potential reviewers / review teams and provide guidance through the general set up of evidence synthesis.

Sharing research data
GRACE initiated a data sharing network together with the EU-project G-TwYST which is performing a 2-year (combined chronic toxicity and carcinogenicity) feeding study with the transgenic NK603 maize) and the French consortium GMO90+ which is performing feeding studies with MON810 and NK603 maize with a specific scope for extended sets of parameters. The portal CADIMA established by GRACE provides a jointly used repository for the primary trial data from the projects. The data are freely accessible for further investigations and evaluations (creative common attribution license). The repository is also open for future project cooperation and data sharing.
The CADIMA database is also hosting complementary survey data from the EU-funded project PreSto GMO EraNet which address former and potential future topics of GMO research.

Stakeholder involvement
In line with the open science strategy GRACE implemented an unique approach of stakeholder involvement not only for discussing project results but also in the research planning (upstream engagement). This strategy was followed for the research on feeding trials as well as for the systematic reviews and evidence maps. Considerable efforts were taken during the project to make detailed information available to stakeholders – in particular in the context of stakeholder consultations. Moreover, by providing responses to their written comments stakeholders could track if and how their comments were considered – or why they were not considered. 500 to 1,300 stakeholders were contacted for each open consultation step. Each stakeholder interested could participate. The only restriction applied to journalists in order to encourage the free exchange of ideas. In the course of the project a total of 143 individual stakeholders from 19 EU Member States, the EU level, Switzerland, Norway, the USA and international organizations, participated in one or several consultations. All workshops had a very good attendance – between 25 and 54 stakeholders. In all consultations all main stakeholder groups (competent authorities, industry, civil society organizations, researcher) where represented and actively contributing. GRACE received and responded to 1,100 written comments. The overall GRACE stakeholder consultation approach with its associated transparency requirements was very positively acknowledged by participants – both stakeholders and project members. Besides contributing to making the science more robust it helped to raise awareness on both sides – on concerns, potential and limitations of the employed methods for risk assessment, and more broadly for the complexity of the research topics as well as needs and constraints of the project. Moreover, it contributed to mutual trust building.
This strategy of intense stakeholder involvement caused high workload for the project team as well as challenging stakeholders’ engagement to its limits. Nevertheless, the intense process showed that high transparency and the supply of detailed data/information facilitates constructive communication in a controversial scientific field.

Responsible Research and Innovation
Project-specific and other more general insights on stakeholder involvement and open science are particularly relevant and topical as they provide practical insights to assist implementing the core concept dimensions of Responsible Research and Innovation, which underlies both the EC Horizon 2020 program and many national programs.
The stakeholder involvement and transparency implemented in GRACE clearly addresses RRI issues. The research topics of GRACE per se address societal concerns, e.g. trust in risk assessment of GM crops and animal welfare. Therefore GRACE may be considered as a reality check for some of the RRI dimensions and their implementation in the context of an EU-funded research project. The project clearly demonstrated the demands of intense stakeholder involvement and increased transparency. Responses to stakeholder requests showed that options to modify or to add additional research activities were limited due to time and financial constraints. Flexibility with the project time schedule was actually not granted by the European Commission for formal reasons. Hence to conform to RRI requirements needs thorough planning to allocate temporal and financial resources in the Grant Agreement, and/or the project management. The Commission needs to allow more flexibility for related RRI issues in the course of a project. Redesigning trials as performed by GRACE required additional time and labour to launch calls for tenders. Shifts in the time schedule combined with shifting major costs create financial challenges if cash flow from interim payments cannot be adjusted during the project. The GRACE team is offering to share its experiences in support of RRI.

General dissemination and communication
The project website as the central public information hub of the project was successfully established. GRACE research activities are well ranked in the Google (and other) search engines so that they show up on the first result page. The ranking has significantly improved during the whole project period. There are about 940 external backlinks found on referring external pages that link to www.grace-fp7.eu. The number of visitors has increased continuously since the launch of the project website. At the end of the reporting period the number of visits stands on average at about 2,000-3,000 (up to 3,500) per month with up to 6,000 page views. The average visit duration (3 min 30 s) indicates a high visit engagement and interest in the provided information. The dissemination of GRACE related information has been enhanced through the use of online-PR portals, other portals and direct mailings to dissemination networks (1,100 contact persons from European consumer organisations, farmer associations, relevant food safety authorities, environmental organisations, and animal welfare organizations), other stakeholder contacts (1,500 contacts to stakeholders from industry, industry associations, scientists, participants of other EU projects, interested persons registered via the project website) and media contacts (about 650 journalists active in the area of GMO products, GMO food/feed, agricultural politics, environmental issues).
A GRACE YouTube channel for presenting GRACE videos has been established. The following set of video productions is being published: Explanation of systematic reviewing, presentations at the GRACE Final Conference, and five video interviews with WP leaders. ARTE TV was supported in producing a video documentation on GMO impact research (documentation of the GRACE Final Conference including video interviews with three GRACE partners).

Major joint dissemination activities with other projects
GRACE partners continued substantial cooperation with other related projects namely G-TwYST (EU, common partners; complementary feeding studies); PreSto GMO EraNet (EU; joint use of CADIMA as info hub); GMO90+ (French consortium; complementary feeding studies); Collaboration of Environmental Evidence - CEE (GB, systematic reviews). Representatives from the EU-funded projects Marlon, Presto GMO ERA-Net, G-TwYST and Res-AGorA actively took part in the GRACE Final Conference to present these complementing projects.

Essential consultations with third parties
• Meetings and consultation with EuropaBio: introduction of the project, introduction of evidence synthesis activities, request for access to use application data from applicants for evidence synthesis
• Meetings and consultation with EFSA: Access to data from applications; workshop on evidence synthesis and further cooperation to explore systematic reviews
• Meetings and consultation with the Collaboration of Environmental Evidence – CEE: joint efforts to promote systematic reviews, Memorandum of Understanding for the use and development of CADIMA web tools, joint workshop on stakeholder involvement in systematic reviews

Conference and workshop sessions actively organized by GRACE Partners
• 5 stakeholder workshops in the project period (in Vienna, Berlin, Brussels)
• GRACE Final Conference, Potsdam, Germany, November 2015
• Session on evidence synthesis at the 13th International Symposium on the Biosafety of Genetically Modified Organisms (ISBGMO), Cape Town, South Africa, November 2014
• Session on evidence synthesis at the IOBC EIGMO workshop, Sofia, Bulgaria, June 2015

Exploitation of results
The portal CADIMA (www.cadima.info) will continue to be operated by the Julius Kühn-Institut after the project ends. It is currently hosting
• the Central Access Database for GMO safety related information sources (free access),
• the Evidence Synthesis Service tools and repository (open access to GRACE review protocols, open access to reviews after publication; access for review teams to evidence synthesis tools after registration; access to reviews from third parties according to the authors licensing conditions),
• the repository for the data from feeding trials with whole GM food/feed (open access to GRACE’s data after publication), and
• the repository of surveys performed by the Project PreSto GMO EraNet (currently access for the project partners only).
All data generated during the feeding trails and the alternative investigations will be published subsequently in 2016. The data will further be shared for joint analyses with the EU-funded project G-TwYST and the French project GMO90+. Open access publications of peer-reviewed papers on the outcomes of the feeding trials detailed raw data will be freely accessible via the portal CADIMA.
A scientific discussion forum has been established in the journal Archives of Toxicology inviting scientific discussions on the outcome and interpretation of feeding studies performed by GRACE, G-TwYST, GMO90+ and related projects as well as of alternative approaches to assess the potential health impacts of products derived from the use of crop genetic improvement technologies.
The systematic reviews and evidence maps on health, environmental and socio-economic impacts prepared for open access publications will become available on the CADIMA website.
Based on the outcome of the evidence map on feeding studies with whole food/feed derived from GM crops the review team decided to proceed with a systematic review on a specific section of that map.
JKI and CEE are organising a training workshop in April 2016 for interested stakeholders/researchers on how to perform evidence synthesis using the tools provided through CADIMA.

List of Websites:
www.grace-fp7.eu;
www.cadima.info;
joachim.schiemann@jki.bund.de;
ralf.wilhelm@jki.bund.de