SUMMARY RISK ASSESSMENT OF GMO PRODUCTS IN THE EUROPEAN UNION- Toxicity assessment, allergenicity assessment and substantial equivalence in practice and proposals for improvement and standardization -The full report is available below or at http://www.umweltbundesamt.at/publikationen/publikationssuche/publikationsdetail/?&pub_id=1531Authors: Armin Spök, Heinz Hofer, Petra Lehner, Rudolf Valenta, Susanne Stirn, Helmut Gaugitsch Year: July 2004, posted on the Austrian Federal Environment Agency in 2005 (the original German language monograph in 3 volumes of 730 pages were published in 2002-2003)Updated and based on 2 research projects conducted by the Umweltbundesamt Wien (Federal Environment Agency of Austria), the Inter-University Research Centre for Technology, Work and Culture (IFZ) in Graz, the ARC Seibersdorf research GmbH, the Research Center for Biotechnology, Society and the Environment (BIOGUM) at the University of Hamburg and a range of experts contracted on a personal basisResearch funded by the Austrian Federal Ministry for Work and Labour, and the Federal Ministry for Health and WomenMonograph funded by the Federal Ministry for Agriculture, Forestry, Environment and Water EconomyGenetically modified plants (GMP) as well as derived feed and food have to undergo a risk assessment prior to market authorisation in the EU. The particular requirements for risk assessment have been and still are particularly contested issues and only recently attempts were begun to specify requirements in greater detail. Against this backdrop the practice of risk assessment was investigated by reviewing both Directive 90/220/EEC and Novel Food dossiers. The focus thereby was on toxicity and allergenicity assessment, on how the concept of substantial equivalence was being put into practice but also included some general aspects of risk assessments.The review revealed a number of shortcomings in both type of dossiers:1. Shortcomings in the overall risk assessment approach: the formal structure is not based on and does not clearly distinguish between exposure assessment and hazard assessment which are both considered necessary to allow for a proper risk assessment. Substantial equivalence plays a key role in both type of applications. In contrast to its conceptual role as a starting point in risk assessment, it rather denotes a terminal stage. The claims of substantial equivalence are frequently based on field trials and compositional analysis that are not properly designed and are often not backed up by throughout and consistently applied statistical analysis. 2. Risk assessments and safety conclusions drawn frequently cannot be entirely verified or even not verified at all on the basis of information presented in the dossiers given the lack of details in the description of tests, approaches, in data display and the tendency not to include full reports. 3. Overall approaches in risk assessment are similar in the dossiers, differences became evident at the level of details, though. These differences, especially between dossiers pertaining the same plant species and/or aiming at similar applications might suggest differences in the soundness of the assessment procedure and clearly point to a lack of details in guidance documents. 4. Safety conclusions are a. often based on indirect evidence and/or assumption based reasoning while direct testing of toxic or allergenic properties is rather limited if conducted at all;b. partly based on methods, approaches and assumptions that are questionable (e. g. homology and in vitro digestibility studies in toxicity assessment; studies and assumptions of the decision tree approach in allergenicity assessment);c. not backed up by throughout compliance in safety relevant studies to a quality assurance system;d. largely focussing on the novel proteins introduced only. 5. Unintended effects of genetic modification are usually not investigated and even dismissed. Such effects are apparently believed to be reflected by conspicuous alterations of either morphological or agronomical properties or in key plant compounds. In compositional analysis, however, significant differences found are disregarded without attempts to verify or further investigate these differences in order to enhance the likelihood to detect unintended secondary effects. Whole-plant feeding studies included in the dossiers are feed conversion studies and cannot be considered toxicity studies. These shortcomings might not only diminish the validity of safety conclusions in scientific risk assessment but also reduce their credibility amongst stakeholders and in the general public. On the other hand EU legislative and policy documents are frequently reiterating the need for a high level of safety. With this in mind, proposals were developed aiming at further improvement and standardisation of risk assessment: 6. Overall structure of risk assessment approaches and dossiers: structure of dossiers and risk assessment should be standardised including dedicated chapters to substantial equivalence, exposure, toxicity assessment, and allergenicity assessment. The role of substantial equivalence for risk assessment should be further clarified. Detailed requirements for field trials, sampling and compositional analysis are proposed – significant differences should at least trigger repetition of the analysis including broadening the range of compounds considered. 7. Dossiers should be "stand-alone" including full reports of all available safety relevant studies, quoted literature, statistical evaluation sheets for compositional analysis, thorough descriptions of methods and procedures applied including the type of statistical analysis. 8. Guidance Documents should be further detailed accordingly. 9. Safety relevant statements should specify the nature of evidence supporting the safety claim, e. g. test results, literature data or anecdotal evidence. In general, direct testing of toxic or allergenic properties should be preferred compared to approaches that rely on indications from indirect testing and assumption based reasoning. In case of toxicity assessment a minimum set of endpoints is proposed that largely resembles usually accepted endpoints in other regulatory contexts. Additional endpoints would depend on the particular exposure (especially in case of placing on the market as GMP) and on the results of preceding tests. In case of allergenicity assessment comparative IgE reactivity studies should be conducted using animal models in order to assess both sensitizing and allergenic properties. Safety relevant studies should generally be conducted according to Good Laboratory Practice. Further research should be conducted to clarify the value of studies such as homology comparisons and in vitro digestibility in toxicity assessment. 10. Testing should be extended to include whole-plant/whole-food testing in both toxicity and allergenicity studies in order to more reliably detect unintended and detrimental effects of genetic modification. The first three proposals and partly also those included in the fourth bullet point are deemed to assist both applicants and reviewers – the former who have to conduct the risk assessments and compile the dossiers and the later who have to evaluate the risk assessments in the process of market authorisation. The other proposals are rather aiming at providing a more appropriate factual basis for safety conclusions to be drawn.Some of these proposals might be immediately acceptable and easily be implemented. In fact, some of the proposals of this study have already been included in most recent guidance documents issued by parallel initiative at the level of the European Commission, international organisations and elsewhere. Others might require further discussion and even to conduct additional studies, for instance the particular minimum set of toxicity endpoints. Some proposals might require the further improvement and validation of testing methods, such as whole-plant toxicity studies or even to further develop novel testing procedures, such as the application of animal models in allergenicity assessment.