Schenkelaars Biotechnology Consultancy,
The Netherlands, June 2001

In commission of the Dutch Foundation ‘Consument & Biotechnologie’

• The assumption that organisms modified by rDNA techniques do not pose unique hazards compared to organisms modified by traditional means forms the conceptual cornerstone of OECD guidelines for safety in biotechnology. This assumption has however been challenged in scientific literature and empirical data are lacking to validate the simple linear model of ‘precise’ genetic engineering.
• In 1993, based on its conceptual cornerstone for safety in biotechnology, OECD introduced the concept of substantial equivalence as a guiding principle in the food safety assessment to detect intended and unintended differences between a GM food (component) and its traditional counterpart. This approach has been developed, because in contrast to many compounds such as pesticides, pharmaceuticals, industrial chemicals and food additives, whole (GM) foods are complex mixtures of compounds characterised by a wide variation in composition and nutritional values. Their safety is therefore difficult to assess by conventional toxicological approaches involving for example animal feeding experiments.
• At the end of the 1990s the concept and its application in regulatory decision-making started to attract considerable criticism, as several authors viewed it as an excuse for not requiring toxicological tests. In 2000 expert meetings convened by OECD and FAO/WHO reviewed the application of the concept of substantial equivalence. These meetings resulted in endorsement of the concept in a general sense. However, reports presented at these meetings indicated that there had been a lack of consistency from case to case in the data provided, even within the same crop species. This led to more sophisticated discussions on data requirements to determine substantial equivalence of a GM food crop or ingredients derived thereof, as well as on the ‘traditional counterpart’ or ‘selected comparator’ to be used for comparison and on methods to generate statistically valid data. The expert meetings further recommended developing a database containing baseline concentrations of plant compounds of potential nutritional or toxicological concerns and knowledge on how concentrations of these compounds may vary depending on the genetic background of the plants and environmental conditions.
• According to the OECD and FAO/WHO expert meetings in 2000, present approaches to detect unintended differences between a GM food crop or component and a ‘selected comparator’ due to genetic modification are based on chemical analysis targeted at specific (known) compounds. Further development and validation of profiling techniques based on genomics, proteomics and metabolomics may increase the potential to detect unintended differences.

• Until 2000, within the framework of Regulation 258/97, the concept of substantial equivalence has triggered a series of regulatory decisions by the European Commission and national competent authorities whether a ‘light’ notification procedure or a ‘heavy’ authorisation procedure had to be followed.
• When Regulation 258/97 came into force in January 1997, an operational definition of the concept of substantial equivalence was not available. Nonetheless, case by case, several GM plants and/or ingredients thereof have been determined as substantially equivalent (except for the modified trait) and notified for food use in the EU. However, according to literature, relevant data about inherent plant toxins and anti-nutrients were often missing or showed significant differences. In addition, data for comparisons showed inconsistency from case to case, even within the same plant species.
• An operational definition of substantial equivalence is still lacking. There is for example no minimum list of macro- and micro-nutrients, inherent plant toxins, anti-nutrients, secondary plant metabolites and allergens known to be associated with a crop species, which should be analysed, for the determination of a GM food crop as substantially equivalent. Further, discussions on valid methods to generate compositional data of a GM food crop and its ‘control’ from field trials and on their statistically analysis have not yet been completed by EC scientific committees and competent authorities of EU member states.

• In practice mainly the UK, Germany and The Netherlands have received requests for notification or authorisation of GM food crops. Opinions of EC scientific committees and assessment reports by the competent authorities of the United Kingdom and The Netherlands on the safety evaluation of GM food crops, including a determination of their substantial equivalence, are publicly disclosed and relatively easily accessible through the Internet. The data submitted by applicants are also made publicly available at governmental libraries in these countries. In Germany, however, the competent authority does not have a mandate to publicly disclose the data submitted and its assessment reports under Regulation 258/97. Its substantial equivalence assessments were based on information submitted for commercial releases within the framework of Directive 90/220. In Germany these 90/220 application files are not made publicly available, whereas at a governmental library in The Netherlands these files are publicly accessible.

• Initial assessments by the German, respectively UK competent authority determined refined oil from Liberator Phoe6/Ac, Falcon GS40/90, MS8xRF3 and Topas 19/2 as substantially equivalent. The UK authority requested the applicant to monitor the seed composition and fatty acid profile of the oil of Topas 19/2 over time, as there was little experience in predicting the effect of genetic drift on the metabolism of any plant, whether GM or conventionally bred. In all these cases, the EC Scientific Committee on Plants also determined (refined oil of) these GM rape plants as substantially equivalent (except for the modified trait).
• However, in all cases differences in composition of the oil and/or the meal (for feed use) between the GM-plant and its non-GM control have been observed by member states. Further, from case to case, there has been a lack of consistency in compositional data submitted on the content of macro- and micronutrients, minerals, vitamins, inherent plant toxins and anti-nutrients. The content of sinapine, an anti-nutrient of rape, has not been determined in all cases. In addition, the design of the field trials, the number of locations and seasons, and the choice of the ‘comparator’ have considerably differed from case to case. It is hardly plausible that the compositional data have been analysed in a statistically sound way.

• The initial assessments for notification of Bt11 silage maize, T25 and MON810 by the UK authority determined that these GM maize plants did not differ in composition from their non-GM controls. As there was little experience in predicting the effect of genetic drift on the metabolism of any line of plants, whether GM or non-GM, the UK authorities asked the applicant to monitor the seed composition and the fatty acid and amino acid profile of the GM maize over time. In 1998 the EC Scientific Committee on Plants determined Bt11 (silage) maize as substantially equivalent, whereas in 2000 this committee did not explicitly reach the conclusion that it should be viewed as substantially equivalent. T25 and MON810 were both determined to be substantially equivalent (except for the modified trait) by the EC Scientific Committee on Plants.
• However, in all these cases of notifications differences in the composition of the GM maize plant and its non-GM control have been observed. Further, from case to case, there has been a lack of consistency in compositional data submitted on the content of macro- and micro-nutrients, minerals, vitamins, inherent plant toxins and anti-nutrients. For example, the content of trypsin inhibitor, an anti-nutrient in maize, has not been determined in all cases. In addition, the design of the field trials, the number of locations and seasons, and the choice of the ‘comparator’ have considerably differed from case to case. It is hardly plausible that the compositional data have been analysed in a statistically sound way.
• The requests for authorisation of Bt11 sweet maize and GA21 have been submitted to the Dutch authorities, which treated the data submitted in a rather critical way. Applicants were urged to provide additional data on the content of five secondary metabolites in the GM maize plants compared to their non-GM controls to underpin the degree of substantial equivalence. Such data were provided, but it did not lead unambiguously to a determination of (the degree of) substantial equivalence of GA21. The Dutch authorities noted that applicants would be helped by concrete guidance concerning the number of samples, locations and years, which would be needed for the quantitative analyses.
• In both these cases of a ‘heavy’ authorisation procedure, several member states raised critical questions on the assessment by the Dutch authorities. In the case of Bt11 sweet maize the applicant has provided a response to these questions. In the case of GA21 it is not clear whether the applicant has provided additional information. In both cases it is unclear how these questions and responses impact regulatory decision-making by the European Commission and national authorities of EU member states, as the EC Scientific Committee on Plants has determined Bt11 (sweet) maize as well as GA21 maize as substantially equivalent to their non-GM controls.
• The request for authorisation of crosses of T25 and MON810 has been submitted to the Dutch authorities, which have not yet completed their initial assessment. The EC Scientific Committee on Plants concluded that T25xMON810 hybrids are substantially equivalent to T25 and MON810 and non-GM maize.

• The applicant sought a (full) safety evaluation of processed products of GM tomato TGT7F. The UK authority concluded that no nutritional and toxicological differences existed between the GM tomato and its control. The UK authority did not explicitly establish the GM tomato as substantially equivalent, whereas the EC Scientific Committee on Plants and the EC Scientific Committee on Food both concluded that the GM tomato is substantially equivalent.
• Compositional data have been obtained from trials during one year. It is not clear whether these data have been analysed in a statistically sound way. Further, data of the GM tomato and its control regarding several inherent tomato toxins, such as tomatidine, aglycone of tomatine, saponines, coumarins, protease-inhibitor and oxalate, were not provided.

• There are methodological limitations for obtaining meaningful information from conventional toxicological studies on whole (GM) foods. Irrespective of the issue whether genetic modification of food crops involves unique risks compared to conventional breeding, the concept of substantial equivalence could be a guiding principle to detect intended and unintended differences between a GM food crop and its non-GM control to address these limitations. Differences detected should then be focus of further nutritional, toxicological and immunological evaluation.
• In the EU the concept of substantial equivalence urgently requires an appropriate, operational definition, in particular for deciding whether a ‘light’ notification procedure could be followed.
• An operational definition of substantial equivalence should include detailed protocols on the design of the field trials for collecting compositional data of a GM food crop and its non-GM reference. It should also include a minimum list of macro- and micro-nutrients, anti-nutrients, inherent plant toxins, secondary metabolites and allergens to be analysed for each crop species. It should further foresee in validated techniques to establish the content of these compounds in the (GM) plants and common methods to statistically analyse the data. Differences in the composition of a GM food crop and its non-GM reference, whether intended or unintended, should then be subject to a further safety assessment.

In the EU Regulation 258/97 on Novel Foods and Novel Ingredients regulates the food use of (ingredients of) genetically modified (GM) plants. This regulation provides for a simplified procedure for foods derived of genetically modified organisms (GMOs) but no longer containing GMOs which are ‘substantially equivalent’ to existing foods. In such cases the companies only have to notify to the Commission when placing the novel food or novel ingredient on the market. The product can then be marketed in the entire EU. If a GM plant (or ingredients derived thereof) is not determined as ‘substantial equivalent’, the regulation foresees a full authorisation procedure. Hence, the concept of ‘substantial equivalence’ plays a decisive role in regulatory decision-making on the food use of (ingredients of) GM plants in the EU.

Moreover, Regulation 258/97 stipulates that no later than five years from the date of entry into force and in the light of the experience gained, the Commission shall forward to the European Parliament and to the Council a report on its implementation. The date of entry was 27 January 1997, which implies that the Commission should forward this report at the latest on 27 January 2002.

Further, the Dutch Foundation ‘Consument en Biotechnologie’ is involved in a project to actively involve consumer organisations in the further development of regulatory policies on genetically modified foods and their safety evaluation in the European Union. The project is an initiative of Consumentenbond, the Dutch Consumers Union, and has received a grant from the European Commission Directorate General Health and Consumer Protection. In the fall of 2001 a workshop will be convened to have an exchange of views on the implementation of Regulation 258/97 between food officers of national consumer organisations and the European consumer organisation BEUC, scientists and representatives of national competent authorities and the European Commission. In addition, BEUC, participating in ENTRANSFOOD, a European research project on the food safety assessment of genetically modified food crops, will be enabled to provide adequate input into this research project.

Within this context the Dutch Foundation ‘Consument en Biotechnologie’ commissioned Schenkelaars Biotechnology Consultancy to prepare an analysis of international and European regulatory discussions on the concept of substantial equivalence. This analysis should also include a set of case studies on how this concept has been applied in notifications and authorisations of (ingredients of) several GM foods crops the EU. The analysis should thereby mainly focus on notifications, as in these cases determination of a GM plant (or ingredient thereof) as substantially equivalent triggers the regulatory decision that it can be placed on the market.

The historical analysis of regulatory and scientific discussions on the concept of substantial equivalence has been based on reports of expert meetings convened from 1990 to 2000 by the Organisation for Economic Co-operation and Development (OECD) and the United Nations Food and Agriculture (FAO) and World Health Organisation (WHO). Reports of expert meetings held in the United States in 1999 and Canada in 2000 are also reviewed. This analysis can be found Chapter 2 of this report. Short summaries of the aforementioned documents can be found in Technical Annex I.

The analysis of the concept of substantial equivalence in relation to European legislation has been based on a review of the European Regulation 258/97, opinions delivered by several scientific committees of the Commission, a paper by the United Kingdom Food Standards Agency, and a review article published by a scientific journal. This analysis can be found in the Chapter 3. Technical Annex II contains short summaries of these documents.

The case studies have been based on various documents obtained from the websites of the European Commission and of national competent authorities of Germany, The Netherlands and the United Kingdom. Several of these documents have been accessed at the libraries of the Dutch Ministry of Public Health, respectively the Ministry of Environment. Additional information has been obtained through personal communication with Mrs Van de Wiel of the Dutch advisory body, with Mrs Schauzu of the German competent authority and with Mr Ball of the United Kingdom advisory body. The analysis of the case studies can be found in Chapter 4. Technical Annex III contains detailed accounts of the findings in each case.

The full references, on which this report has been based, can only be found in the technical annexes.

The focus of this study is on the ways the concept of substantial equivalence has been discussed by various international and European bodies and on its application by the European Commission and national competent authorities.

Approaches applied by applicants and regulatory authorities to assess potential toxicological and immunological impacts of newly introduced DNA or protein(s) in a GM food crop are beyond the scope of this study. Main reason is that there seems to be general scientific and regulatory consensus that newly introduced DNA or protein(s) are not to be considered as substantially equivalent and therefore require particular attention per se in the food safety assessment of (ingredients of) GM food crops.

Hence, this study does not evaluate the full food safety assessments of GM food crops as the European Commission and national competent authorities have carried these out.

In the beginning of the 1970s the advent of recombinant DNA (rDNA) techniques, which enabled splicing DNA molecules and reintroducing them into living organisms, almost immediately prompted the first discussions among scientists on whether the use of these rDNA techniques would go associated with unique hazards. It also led to the first discussions among governmental policymakers whether application of rDNA techniques should be regulated. At that time application of these rDNA techniques was limited to laboratories and the first safety guidelines were therefore developed for the use of rDNA organisms in confinement. About ten years later these techniques were also increasingly applied to develop novel organisms for experiments and usage outside laboratories. In 1983 this led the Organisation for Economic Co-operation and Development (OECD) to develop in general safety guidelines for the use genetically modified organisms (GMOs) in industry, agriculture and the environment, which were published in 1987. An important principle was to focus safety concerns on whether organisms modified by rDNA techniques pose an ‘incremental’ risk to organisms modified by conventional means. Although application of rDNA techniques might result in organisms with a combination of traits not observed in nature, genetic changes from rDNA techniques would often have a greater predictability compared to traditional means because of the greater precision of rDNA techniques. The expectation therefore was that any risk associated with application of GMOs might be assessed in generally the same way as those associated with non-GMOs.

This conceptual cornerstone of the OECD safety guidelines, upon which regulations for safety in biotechnology of most industrialised countries have been based, has however not been without any scientific controversy. In scientific literature several authors have argued that unique hazards do exist in the movement of genes between unrelated organisms by applying rDNA techniques. Also a few scientists of the US Food and Drug Administration (FDA) have cautioned this agency in the beginning of the 1990s that foods produced through rDNA technology entail different risks than do their conventional counterparts but their view has been disregarded. Moreover, two expert committees, one convened by the US National Research Council in 1999 and the other one convened by the Royal Society of Canada in 2000, concluded that empirical data were lacking to validate the simple linear model of ‘precise’ genetic engineering.

Three years after the publication of general safety guidelines for the use of GMOs, in 1990, the OECD established a working group for developing scientific principles on the safe use of foods derived by modern biotechnology. In 1993 this working group recommended to base the scientific approach to the safety evaluation of a new food or food component on the comparison with traditional foods that have a safe history of use. This approach was in turn based on the concept of substantial equivalence, which articulated procedures used in the past, albeit intuitively, for accepting new foods. Use of this approach should lead to reasonable certainty that no harm will result from the intended uses under anticipated conditions of consumption, as in the working group’s view foods derived by modern biotechnology were inherently not less safe than their origin counterparts. A fundamental change in the safety assessment of such foods was therefore not considered as necessary, nor should a different standard of safety be required. The principle was set out that if a modified food or food component would be determined to be substantially equivalent to an existing food, it could be treated in the same manner. No additional safety concerns would be expected. However, when a food or food component would not be determined to be substantially equivalent, the identified differences should be the focus of further evaluations. A number of theoretical case studies illustrated how ‘substantial equivalence’ might be applied.

An important rationale for developing this approach to the safety assessment of GM foods was that in contrast to many compounds such as pesticides, pharmaceuticals, industrial chemicals and food additives, whole (GM) foods are complex mixtures of compounds characterised by a wide variation in composition and nutritional values. To the opinion of the OECD working group, there were serious practical and methodological difficulties of obtaining meaningful information from conventional toxicology studies on the safety of whole foods. These difficulties had for example also become apparent when animal feeding studies had been used to assess the safety of irradiated foods.

In 1996 the United Nations Food and Agriculture Organisation (FAO) and World Health Organisation (WHO) also convened an expert consultation to address approaches to assess the safety of GM foods. With reference to the work by OECD, this consultation recommended that the concept of substantial equivalence was an important component in the safety assessment of foods and food ingredients derived from GM plants.

In October 1999 the scientific journal Nature published a critical view on the concept of substantial equivalence, which attracted wide attention. The authors argued that the degree of difference between a natural food and its GM alternative before its ‘substance’ ceases to be acceptably ‘equivalent’ had not been defined, nor had legislators agreed an exact definition. The authors felt that the vagueness of the concept might make it useful to industry but unacceptable to the consumer. The concept was considered as inherently anti-scientific because it was created primarily to provide an excuse for not requiring toxicological tests. Although the authors acknowledged that biochemical and toxicological tests and their interpretation are notoriously problematic and contested, and slow and expensive, they could provide information vital to consumer protection. Companies should have been required to conduct a range of toxicological tests to set ‘acceptable daily intakes’.

The OECD however argued that such classical toxicological tests might also well reveal that traditional foods often contain natural toxicants and/or anti-nutritional substances with undesirable effects in animal experiments. It would therefore be unreasonable and unscientific to expect an absolute standard of safety for GM foods, which could be demonstrated through classical toxicological tests. The important starting point in the safety assessment should therefore be a comparison between the new GM food and its traditional counterpart. However, application of substantial equivalence was not in itself to be viewed as a substitute for a safety assessment. The concept had been conceived as a guiding principle, which was intended to be a tool for those engaged in food safety assessments of GM foods to detect intended and unintended differences between a GM food or food component and its ‘traditional counterpart’ or ‘selected comparator’. The OECD also pointed out that there was work underway to identify the critical nutrients and toxicants found in major crops, which would be an important focus of a demonstration of substantial equivalence.

Since 1995 regulators from countries in North America, Europe and Asia had gained practical experience in applying the concept of substantial equivalence but there had been no universal consensus on its application. In 2000 a series of expert meetings convened by OECD, respectively by FAO/WHO reviewed the concept, its limitations and its role in regulatory decision-making. In a general sense all these expert meetings endorsed the concept of substantial equivalence as a tool in the safety assessment of GM foods and GM food ingredients.

Reports presented at these meetings however indicated that the comparison principle of substantial equivalence had normally been used in regulatory decision-making but there had been a lack of consistency from case to case in the data provided, even within the same crop species. Consequently, discussions on data requirements became more sophisticated. For example, when determining substantial equivalence of a GM food crop and/or its ingredients, one of the expert meetings recommended that the ‘selected comparator’ should be ideally the near isogenic parental lines grown under identical conditions. It further recommended developing guidelines for appropriate design of field trials with GM food crops and guidelines for key substances, nutrients and data needed for assessment of substantial equivalence. It also recommended that data for comparison should be obtained using validated methods and analysed using appropriate statistical techniques.

Another aspect addressed by these meetings was a need for a data base containing baseline concentrations of plant compounds of potential nutritional or toxicological concerns and knowledge on how concentrations of these compounds may vary depending on the genetic background of the plants and environmental conditions. In addition, these expert meetings indicated that present approaches to assess possible unintended effects are based on chemical analyses targeted at specific (known) compounds. In order to increase the probability of detecting unintended effects, a further development and validation of profiling techniques based on genomics, proteomics and metabolomics was generally recommended to detect differences in a more extensive way than targeted chemical analysis.

• The assumption that organisms modified by rDNA techniques do not pose unique hazards compared to organisms modified by traditional means forms the conceptual cornerstone of OECD guidelines for safety in biotechnology. This assumption has however been challenged in scientific literature and empirical data are lacking to validate the simple linear model of ‘precise’ genetic engineering.
• In 1993, based on its conceptual cornerstone for safety in biotechnology, OECD introduced the concept of substantial equivalence as a guiding principle in the food safety assessment to detect intended and unintended differences between a GM food (component) and its traditional counterpart. This approach has been developed, because in contrast to many compounds such as pesticides, pharmaceuticals, industrial chemicals and food additives, whole (GM) foods are complex mixtures of compounds characterised by a wide variation in composition and nutritional values. Their safety is therefore difficult to assess by conventional toxicological approaches involving for example animal feeding experiments.
• At the end of the 1990s the concept and its application in regulatory decision-making started to attract considerable criticism, as several authors viewed it as an excuse for not requiring toxicological tests. In 2000 expert meetings convened by OECD and FAO/WHO reviewed the application of the concept of substantial equivalence. These meetings resulted in endorsement of the concept in a general sense. However, reports presented at these meetings indicated that there had been a lack of consistency from case to case in the data provided, even within the same crop species. This led to more sophisticated discussions on data requirements to determine substantial equivalence of a GM food crop or ingredients derived thereof, as well as on the ‘traditional counterpart’ or ‘selected comparator’ to be used for comparison and on methods to generate statistically valid data. The expert meetings further recommended developing a database containing baseline concentrations of plant compounds of potential nutritional or toxicological concerns and knowledge on how concentrations of these compounds may vary depending on the genetic background of the plants and environmental conditions.
• According to the OECD and FAO/WHO expert meetings in 2000, present approaches to detect unintended differences between a GM food crop or component and a ‘selected comparator’ due to genetic modification are based on chemical analysis targeted at specific (known) compounds. Further development and validation of profiling techniques based on genomics, proteomics and metabolomics may increase the potential to detect unintended differences.

On 25 January 1997 Regulation 258/97 on Novel Foods and Novel Foods Ingredients came into force in the EU. Food or food components derived of GM plants also fall under this regulation, which foresees a simplified notification procedure if such GM food or food component is determined as ‘substantially equivalent’ to an existing food with respect to composition, nutritional value and metabolism. In other cases a full authorisation procedure must be followed.

On 29 July 1997 the Scientific Committee for Food (SCF) of the European Commission published its recommendations concerning the scientific aspects and the information necessary to support an application of a novel food or novel food ingredient. With reference to recommendations by OECD and FAO/WHO the SCF reminded that determination of substantial equivalence is not a safety or nutritional assessment in itself but an approach to compare a novel food with its conventional counterpart. Analytical studies were viewed of crucial importance for the establishment of substantial equivalence as well as for nutritional and toxicological assessments. Methods had to be standardised and validated. SCF further recommended to focus investigations especially on the content of critical macro- and micro-nutrients, any critical toxicants and anti-nutritional factors which might be either inherently present or due to the genetic modification.

On 28 December 1998 the Scientific Committee on Plants (SCP) of the European Commission published a guidance document to facilitate applicants in the preparation of GM plant dossier. One of its recommendations was that data should be obtained from a valid comparison of GM plants and a conventional, preferably isogenic line, based on samples from at least two seasons from a number and variety of geographical locations and accompanied by an appropriate statistical analysis. In May 1999 the scientific advisory body to the UK regulatory authority also presented a paper with considerations for statistically valid data to the European Commission to promote discussion.

In October 2000 the Scientific Steering Committee of the European Commission endorsed substantial equivalence as a starting point to guide the safety evaluation. It noted that several scientific committees had used the concept on a case by case base and therefore needed an operational definition of the concept. Further, the routine analytical procedures in use also needed to be standardised in terms of sampling and extraction procedures, validation of profiling methods and in bio-informatics.

Finally, in 2001 the Scientific Committee on Plants (SCP) began drafting a guidance document to facilitate applicants in the preparation of GM plant dossiers for an authorisation of a GM plant release (Directive 90/200 and Directive 2001/18) and subsequently for a GM food (Regulation 258/97) (and/or feed) authorisation. The opinion would fall under the remit of three scientific committees: Scientific Committee on Food (SCF), Scientific Committee on Animal Nutrition (SCAN) and SCP. The guidance document should in principle be adopted by the Scientific Steering Committee (SSC). The document by SCP would provide guidance for the molecular characterisation of the gene construct, for the establishment of substantial equivalence and for performing the safety assessment, the nutritional assessment and the environmental assessment.

Since Regulation 258/97 came into force in January 1997, an operational definition of the concept of substantial equivalence for decision-making has been and still is lacking. Nonetheless, since 1995 several GM plants and/or ingredients thereof have been determined as substantially equivalent and notified for food use in the EU. In several cases of authorisation the concept of substantial equivalence also played an important role in their safety assessment.

In 2000 a scientific review article (Kovak 2000), based on an analysis of data submitted for notification of the food use of several GM plants and/or ingredients thereof in the US and EU until 1998, generally supported the use of substantial equivalence. Nonetheless, this review article revealed that relevant data about inherent plant toxins and anti-nutrients were often missing or showed significant differences. Data for comparisons showed inconsistency from case to case. It was therefore recommended to draft a minimum list of macro- and micro-nutrients, inherent plant toxins, anti-nutrients, secondary plant metabolites and allergens known to be associated with a crop species. In one case significant differences between a GM oilseed rape plant and its control in the values for an inherent plant toxin had been observed but its content varied more between the different locations than between different GM and non-GM oilseed rape lines at a given location. According to the applicant, it was generally accepted that commercial food derived of plants exhibit considerable variability in its composition. In this case the differences observed were more the result of the interaction of the genotype with the environment rather than the result of the insertion of specific genes into the plant genome. Since this might not be scientifically justified, the authors raised question such as: How much is the accepted range of variability? What are the regulations for variations above the statistical limits? And what variations might be dangerous? It was also pointed out that literature data sometime show wide variations in typical plant components and may not be directly comparable due to differences in analytical methods or sample preparations. Further, literature data are often relatively old and may not encompass the compositional variables of modern crop varieties. Another problem for obtaining appropriate data observed by the authors was that it is technically not always possible to use authentic isogenic lines to compare the GM lines with.

Another scientific article (Noteborn 2000) described a recently developed chemical fingerprinting methodology that could contribute to the demonstration of substantial equivalence by its ability to compare compositional changes with regard to a couple of hundred low-molecular compounds between a GM food crop and related non-GM reference lines. The authors also proposed a hierarchical approach by comparing the chemical fingerprints of the GM plant to those of: 1) isogenic parental or closely related lines bred at identical and multiple sites; 2) extended ranges of commercial varieties of that plant; and 3) downstream processing effects. Such an approach was recommended to assess the likelihood that the statistical differences in a GM crop plant may be false positives due to chance alone, or arose from natural genetic and/or physiological variations.

• Until 2000, within the framework of Regulation 258/97, the concept of substantial equivalence has triggered a series of regulatory decisions by the European Commission and national competent authorities whether a ‘light’ notification procedure or a ‘heavy’ authorisation procedure had to be followed.
• When Regulation 258/97 came into force in January 1997, an operational definition of the concept of substantial equivalence was not available. Nonetheless, case by case, several GM plants and/or ingredients thereof have been determined as substantially equivalent (except for the modified trait) and notified for food use in the EU. In several cases of authorisation the concept of substantial equivalence also played an important role in their safety assessment. However, according to literature, relevant data about inherent plant toxins and anti-nutrients were often missing or showed significant differences. In addition, data for comparisons showed inconsistency from case to case, even within the same plant species.
• An operational definition of substantial equivalence is still lacking. There is for example no minimum list of macro- and micro-nutrients, inherent plant toxins, anti-nutrients, secondary plant metabolites and allergens known to be associated with a crop species, which should be analysed, for the determination of a GM food crop as substantially equivalent. Further, discussions on valid methods to generate compositional data of a GM food crop and its ‘control’ from field trials and on their statistically analysis have not yet been completed by EC scientific committees and competent authorities of EU member states.
• Recent advances in chemical fingerprinting of a couple of hundred low-molecular compounds in complex plant matrices could further contribute to detect unintended changes in the composition of a GM food crop compared to its non-GM reference lines.

To analyse the application of substantial equivalence in notification and authorisation procedures in the EU, detailed case studies have been prepared on:

• Notification of refined oil of GM rape Liberator Phoe6/Ac.
• Notification of refined oil of GM rape Falcon GS40/90.
• Notification of refined oil of GM rape MS8xRF3.
• Notification of refined oil of GM rape Topas 19/2.
• Notification of processed products of GM Bt11 silage maize.
• Request for authorisation of processed products of GM Bt11 sweet maize.
• Request for authorisation of processed products of GM maize GA21.
• Notification of processed products of GM T25 maize.
• Notification of processed products of GM MON810 maize.
• Request for authorisation of processed products of conventional crosses of GM maize T25 and GM maize MON810.
• Authorisation of processed products of GM tomato TGT7F in the UK.
• Request for authorisation of processed products of GM tomato TGT7F in the EU.

The findings described below are based on the detailed accounts of each case presented in Technical Annex III. The information for these accounts has been obtained in several manners.

Documents with opinions on substantial equivalence delivered by the Scientific Committee for Foods and the Scientific Committee on Plants of the European Commission were accessible at the website of the European Commission. These documents also contained a short summary of the data submitted. The original data sets, on which the substantial equivalence assessments have been based, were however not publicly available through the Internet.

Initial assessment reports on substantial equivalence of (ingredients of) GM plants were available at the website of the competent authority of the United Kingdom, or have been obtained through correspondence. The data sets submitted can in principle be accessed at the British library.

The German authority indicated that it had no mandate to make its initial assessment reports on notifications of (ingredients of) GM plants made publicly available. It further indicated that no separate notification files for the food use of refined oil of these three GM oilseed rape plants had been submitted. Its substantial equivalence assessments were based on information submitted in the application of Liberator Phoe6/Ac, Falcon GS40/90 and MS8xRF3 for commercial release within the framework of Directive 90/220. These application files are neither publicly disclosed by the German authorities.

In the Netherlands, similar to the UK, initial assessment reports by the Dutch authorities are made publicly available at the website of the scientific advisory body to the Ministry of Public Health. Further, upon request, the data sets submitted and correspondence with the European Commission, national authorities of other EU member states and the applicants are also made available. Noteworthy, the Dutch authorities also disclose, upon request, the initial assessment reports by the German authorities on Liberator Phoe6/Ac, Falcon GS40/90 and MS8xRF3.

Moreover, the Ministry of Environment, the Dutch authority for Directive 90/220 on the deliberate release into the environment of GMOs, has made all application files for experimental and commercial releases of GMOs publicly accessible at its library with relatively little information kept confidential. Information on the data submitted in the case of Liberator Phoe6/Ac, Falcon GS40/90 and MS8xRF3 could therefore collected from these application files.

It should also be noted that collecting the information for the case studies on substantial equivalence assessments of (ingredients of) GM food crops by various scientific committees of the European Commission and EU member states required substantial efforts. It should also be noted that original data sets submitted to determine substantial equivalence have until now not been routinely exchanged between the member state, which drafted the initial assessment report, and other member states and the European Commission. In addition, opinions of scientific committees of the European Commission and initial assessment reports by national authorities of EU member states, which could be accessed, all contained different kinds of summaries of information and data, on which regulatory decisions have been based. There is however a lack of transparency about the way these authorities have summarised the information and data submitted in each case.

In summary, opinions of EC scientific committees and assessment reports by the competent authorities of the United Kingdom and The Netherlands on the safety evaluation of GM food crops, including a determination of their substantial equivalence, are publicly disclosed and relatively easily accessible through the Internet. The data submitted by applicants are also made publicly available at governmental libraries in these countries. In Germany, however, the competent authority does not have a mandate to publicly disclose the data submitted and its assessment reports under Regulation 258/97. It turned out that its substantial equivalence assessments were based on information submitted in the application of Liberator Phoe6/Ac, Falcon GS40/90 and MS8xRF3 for commercial release within the framework of Directive 90/220. In Germany these application files are not made publicly available, whereas at a governmental library in The Netherlands these files are publicly accessible.

The application file, which has been submitted under Directive 90/220 in Germany in 1998, indicated that the data for the comparison of the contents of micro-nutrients (sterols and tocopherols) and inherent plant toxins (erucic acid) of refined oil of Liberator Phoe6/Ac plants and their controls have been obtained from pooled samples of the one year’s harvest from two different locations. Moreover, data submitted on the content of oil, protein, fatty acids, amino acids, erucic acid and glucosinolates have neither been analysed in a statistically sound way. In desolventising meal of Liberator Phoe6/Ac plants, which might be used as animal feed, the Dutch authority observed an increase of 20 to 25% of glucosinolates.

In 1999 the German authority determined refined oil of Liberator Phoe6/Ac substantially equivalent to that of its parental line Liberator regarding its nutritional value, intended use, composition and the level of undesirable substances. Protein content of the refined oil had been established at 0.01 mg/l based on information submitted in other, similar notifications of refined oil of Falcon GS40/90, Topas 19/2 and MS8xRF3. In June 2000, after questions raised by several member states, the applicant provided additional information on the three different methods that had been used for the analysis of the protein content of refined oil in these three different notifications. This information confirmed that no modified DNA or modified protein was present in refined oil of these three GM oilseed rape plants.

In November 2000 the EC Scientific Committee on Plants also concluded that GM oilseed rape varieties derived of transformation event Liberator Phoe6/Ac are substantially equivalent to their conventional counterparts except for the introduced traits.

The application file, which has been submitted under Directive 90/220 in Germany in 1996, contained a review of the compositional data by the Dutch research institute RIKILT dated on August 1999. In the view of RIKILT, the application had in first instance lacked a statistical analysis of the data on the oil, protein, fatty acid, and amino acid composition of seeds of GM oilseed rape plants. The application in first instance also lacked data on minerals or ash, vitamins (especially vitamin E) and data on the content of glucosinolates in seed and meal and on the content of sterols in the oil. The applicant subsequently provided additional information. Oil content of GM seeds was slightly but not consistently lower than in non-GM seeds, whereas in meal (for feed use) larger differences were found. The contents of individual glucosinolates in GM meal were lower in non-GM meal in most cases. RIKILT’s final conclusion was that the GM seeds did not differ substantially from non-GM seeds except for the presence of a modified protein.

In July 1998 the EC Scientific Committee on Plants concluded that GM oilseed rape plants derived of Falcon GS40/90 are substantially equivalent to their unmodified counterparts except for the inserted traits. One year later, in July 1999, the German authority determined refined oil of Liberator Phoe6/Ac substantially equivalent to that of its ‘comparator’ regarding its nutritional value, intended use, composition and the level of undesirable substances. Protein content of the refined oil had been established at 0.01 mg/l like in other, similar notifications of refined oil of Topas 19/2 and MS8xRF3. In June 2000, after questions raised by several member states, the applicant provided additional information on the three different methods that had been used for the analysis of the protein contents of refined oil in these three different notifications. This information confirmed that no modified DNA or modified protein was present in refined oil of these three GM oilseed rape plants.

The application file, which has been submitted under Directive 90/220 in Belgium in 1996, contained data on the content of oil, fibre, vitamins, minerals, glucosinolates and amino acids in desolventising meal of GM plants, control-plants and commercial varieties, and on the content of chlorophyll, tocopherols, sterols, minerals, fatty acids and erucic acid in processed oil. Samples were pooled from the harvest of one year at many locations. The data were not statistically analysed. In one case the applicant observed significant differences in levels of glucosinolates between RF3 plants and their controls. The applicant argued that these differences were due to environmental conditions. The Belgian authority concluded that MS8xRF3 is substantially equivalent.

Three years later, in November 1999, the German authority for Regulation 258/97 determined refined oil of MS8xRF3 substantially equivalent to that of its comparator regarding its nutritional value, intended use, composition and the level of undesirable substances. Protein content of the refined oil had been established at 0.01 mg/l like in other, similar notifications of refined oil of Topas 19/2 and Falcon GS40/90. In June 2000, after questions raised by several member states, the applicant provided additional information on the three different methods that had been used for the analysis of the protein contents of refined oil in these three different notifications. This information confirmed that no modified DNA or modified protein was present in refined oil of these three GM oilseed rape plants.

In June 1997, based on a compositional analysis of the content of oil, fatty acid, erucic acid and glucosinolates, the UK ACNFP concluded that the refined oil from Topas 19/2 did not differ in composition from conventional rape. However, as there was little experience in predicting the effect of genetic drift on the metabolism of any line of plants, whether GM or non-GM, the UK authorities asked the applicant to monitor the seed composition and the fatty acid profile of GM swede rape over time.

In February 1998 the EC Scientific Commmittee on Plants concluded that refined products from plants derived from Topas 19/2 would be safe for food use on the basis of its substantial equivalence. The data submitted concerned the content of oils, fatty acids, erucic acid and glucosinolates and were obtained from the harvest of trials at a number of locations in several successive years.

• Initial assessments by the German, respectively UK competent authority determined refined oil from Liberator Phoe6/Ac, Falcon GS40/90, MS8xRF3 and Topas 19/2 as substantially equivalent. The UK authority requested the applicant to monitor the seed composition and fatty acid profile of the oil of Topas 19/2 over time, as there was little experience in predicting the effect of genetic drift on the metabolism of any plant, whether GM or conventionally bred. In all these cases, the EC Scientific Committee on Plants also determined (refined oil of) these GM rape plants as substantially equivalent (except for the modified trait).
• However, in all cases differences in composition of the oil and/or the meal (for feed use) between the GM-plant and its non-GM control have been observed by member states. Further, from case to case, there has been a lack of consistency in compositional data submitted on the content of macro- and micronutrients, minerals, vitamins, inherent plant toxins and anti-nutrients. The content of sinapine, an anti-nutrient of rape, has not been determined in all cases. In addition, the design of the field trials, the number of locations and seasons, and the choice of the ‘comparator’ have considerably differed from case to case. It is hardly plausible that the compositional data have been analysed in a statistically sound way.

In February 1998 the UK authority notified processed products of Bt11 silage maize (used for feed) to the European Commission. Its initial assessment report was based on data from the compositional analyses with regard to starch, protein, oil, fibre, fatty acids and amino acids. Although statistically significant differences in protein content were found between some GM hybrids and their controls, the conclusion was that Bt11 silage maize lines would not differ in composition from conventionally bred maize. However, as there was little experience in predicting the effect of genetic drift on the metabolism of any line of plants, whether GM or non-GM, the UK authorities asked the applicant to monitor the seed composition and the fatty acid and amino acid profile of the GM maize over time.

In February 1998, based on data on the content of oil, fatty acids, fibre, starch and amino acids obtained through compositional analysis of seeds from trials at a number of locations, the EC Scientific Committee on Plants concluded that products of imported Bt11 maize are substantially equivalent.

In August 2000 the Italian authority informed the European Commission that it considered Bt11, MON810 and T25 not to be substantially equivalent within the meaning of Regulation 258/97 and should therefore be submitted for a full safety evaluation. According to Italy, the original application lacked PCR data to detect modified DNA in products of these GM maize plants. In September 2000 the EC Scientific Committee on Food however viewed that there were no scientific grounds for considering that the use of these novel foods would endanger human health.

Three months later, in November 2000, the EC Scientific Committee on Plants delivered its opinion on Bt11 maize. It noted several (minor) differences. In its opinion the SCP did not explicitly determine Bt11 maize as substantially equivalent.

In April 1998 the Dutch authority received a request for an authorisation of fresh and processed kernels of Bt11 sweet maize (used for food). The applicant provided data on the content of starch, protein, oil, fibre, fatty acids and amino acids. The Dutch authority however wished additional data on the contents of the five secondary metabolites to underpin the degree of substantial equivalence: furfural, p-coumaric acid, ferulin acid, raffinose and phytic acid. These secondary metabolites were selected because their concentration range in maize was known and because furfural’s potential carcinogenicity. The company provided these data in January 2000. In April 2000 the Dutch authority issued its initial assessment report, which led to several questions from other member states. In November 2000 the company provided a response to these questions.

The initial assessment by the Dutch authority treated the data submitted in a rather critical way. It for example observed significant differences with regard to some nutrients. It also expressed the view that the study on the contents of minerals and vitamins should not have pooled the data obtained from several locations. The analysis of contents of the five secondary metabolites was based on samples taken from three Bt11 hybrids and their controls grown at one location. No statistically significant differences have been found. In its initial assessment report the Dutch authority did not unambiguously establish the (degree of) substantial equivalence of Bt11.

Several member states argued for several reasons that the data submitted were insufficient to establish the substantial equivalence of Bt11. In addition, several member states viewed the statistical analysis of the data inadequate. In its response the company acknowledged that the use of various hybrids, different experimental designs and different analytical methods makes a direct comparison of the numerous compositional analyses difficult. In the view of the applicant, this was due to the fact that the requirements for compositional analyses evolved between 1995 and 2000. In its response the company also provided an extensive justification for the choice of compounds, which have been analysed.

At the moment, it is not clear whether the applicant’s response satisfies the other member states, nor is it clear how this impacts regulatory decision-making on the food use of Bt11 sweet maize by the European Commission and the national authorities of EU member states.

In July 1998 the applicant requested the Dutch authority to authorise GA21 for food use and provided data on Ash, calcium, carbohydrates, acid detergent fibre ADF, neutral detergent fibre NDF, moisture, phosphorous, protein, fat, fatty acids, amino acids, trypsin inhibitors, phytic acid, and vitamin E.

In February 1999 the applicant was requested to provide additional information about the contents of five secondary metabolites furfural, p-coumaric acid, ferulin acid, raffinose and phytic acid. In March 1999 the company argued that it would not be relevant to investigate the levels of these secondary plant metabolites, as maize has a safe history of use. In June 1999 the Dutch authority reiterated its request for additional information, which was provided by the company in October 1999. In December 1999 the Dutch issues its initial assessment report.

The initial assessment by the Dutch authority treated the data submitted in a rather critical way. It noted for example that the level of p-coumarin acid was significantly higher in one of the GA21 hybrids. The applicant argued that this difference was small, fell within the range of values for other commercial maize varieties and was unlikely to be biologically significant. The Dutch authority also noted that applicants would be helped by concrete guidance concerning the number of samples, locations and years, which would be needed for the quantitative analyses. Nonetheless, on the basis of these data (and on data with respect to the newly introduced DNA and protein) submitted, the Dutch authority concluded that consumption of GA21 maize is as safe for humans as non-GM maize. But the Dutch authority did not unambiguously establish the (degree of) substantial equivalence of GA21.

Several member states argued for several reasons that the data submitted were however insufficient to establish the substantial equivalence of GA21. In addition, several member states viewed the statistical analysis of the data inadequate. At the moment, it is not clear whether the questions raised by other member states have been responded, nor is it clear how this impacts regulatory decision-making on the food use of GA21 maize by the European Commission and the national authorities of EU member states.

In September 2000 the EC Scientific Committee on Plants determined that GA21 maize is substantially equivalent to its conventional counterparts (except for the modified trait).

In January 1998, on the basis of data on the contents of fatty acids, protein, amino acids, crude fibre, ash, phytate and moisture, the UK authority concluded that there had been no unintentional changes in the composition of T25 maize. But for some parameters variation between the GM and non-GM lines had been observed. Further, as there was little experience in predicting the effect of genetic drift on the metabolism of any line of plants, whether GM or non-GM, the UK authority asked the applicant that the seed composition and the fatty acid and amino acids profiles of T25 maize lines should be monitored over time.

According to the EC Scientific Committee on Plants in February 1998, integration and expression of the gene did not seem to cause any negative pleiotropic effects. The nutritional composition and the content of anti-nutrients in T25 were within the normal range. It concluded that T25 is substantially equivalent to its non-GM counterpart except for the introduced trait.

In August 2000 the Italian authority informed the European Commission that it considered Bt11, MON810 and T25 not substantially equivalent within the meaning of Regulation 258/97 and should therefore be submitted for a full safety evaluation. According to Italy, the original application lacked PCR data to detect modified DNA in products of these GM maize plants. In September 2000 the EC Scientific Committee on Food however viewed that there were no scientific grounds for considering that the use of these novel foods would endanger human health.

In December 1997 the UK authority notified processed products of MON810 to the European Commission. The applicant provided data on the content of fatty acid, protein, amino acid, crude fibre, ash and moisture content of grain and forage GM and non-GM maize grown in agronomic fields tests in Europe and in the USA. According to ACNFP, there had been no unintentional changes in the composition of the grain from the GM plants or in the plants themselves as a result of the genetic modification. It concluded that the composition of the GM maize does not differ from that of conventionally bred maize. However, there was little experience in predicting the effect of genetic drift on the metabolism of any line of plants, whether genetically modified or conventionally bred. Therefore ACNFP asked the applicant to monitor the seed composition, including amino acid and fatty acid profiles of the oil from MON810 over time.

In February 1998, based on compositional analysis with regard to fatty acids, protein, amino acids, crude fibre, ash, phytic acid and moisture, the EC Scientific Committee on Plants determined MON810 to be substantial equivalent to non-GM maize except for the introduced traits.

In August 2000 the Italian authority informed the European Commission that it considered Bt11, MON810 and T25 not substantially equivalent within the meaning of Regulation 258/97 and should therefore be submitted for a full safety evaluation. According to Italy, the original application lacked PCR data to detect modified DNA in products of these GM maize plants. In September 2000 the EC Scientific Committee on Food however viewed that there were no scientific grounds for considering that the use of these novel foods would endanger human health.

In 1999 the UK authority received an application for a maize hybrid derived from conventional crosses of two GM maize lines approved under Directive 90/220, probably T25xMON810. According to the UK authority, the applicant had heavily relied on the fact that these two GM parents had already been approved, so that very little information on the new hybrid was provided. The assumption that this hybrid would exhibit the same properties as the two GM parent lines was challenged. In particular, the possibility of unintended effects arising from insertion events brought about by the crossing could not be ruled out. At this stage such crosses would need to be assessed on a case by case basis. The UK authority also noted that in other countries, such as the USA, clearance of a GM line also covered further crosses with approved GM lines. It seems however that the application has been withdrawn from the UK.

In April 2000 a request for authorisation of processed products of conventional crosses between T25 and MON 810 maize hybrids was submitted in The Netherlands. The Dutch authority has not yet completed its initial assessment report.

In June 2000, on the basis of compositional analysis of the content of protein, fibre, fat, moisture, ash, fatty acids, amino acids, Ca, K, Mg, Mn, Na, P. folic acid, thiamine, riboflaving, vitamin E, phytic acid and trypsin inhibitors, the EC Scientific Committee on Plants noted significant differences for grain ash, stearic acid, linolenic acid, arginine and riboflavin between T25xMON810 and its control but their levels were within ranges published for maize. It concluded that T25xMON810 was substantially equivalent to its parental lines T25 and MON810.

• The initial assessments for notification of Bt11 silage maize, T25 and MON810 by the UK authority determined that these GM maize plants did not differ in composition from their non-GM controls. As there was little experience in predicting the effect of genetic drift on the metabolism of any line of plants, whether GM or non-GM, the UK authorities asked the applicant to monitor the seed composition and the fatty acid and amino acid profile of the GM maize over time. In 1998 the EC Scientific Committee on Plants determined that Bt11 (silage) maize as substantially equivalent, whereas in 2000 this committee did not explicitly reach the conclusion that it should be viewed as substantially equivalent. T25 and MON810 were both determined to be substantially equivalent (except for the modified trait) by the EC Scientific Committee on Plants.
• However, in all these cases of notifications differences in the composition of the GM maize plant and its non-GM control have been observed. Further, from case to case, there has been a lack of consistency in compositional data submitted on the content of macro- and micro-nutrients, minerals, vitamins, inherent plant toxins and anti-nutrients. For example, the content of trypsin inhibitor, an anti-nutrient in maize, has not been determined in all cases. In addition, the design of the field trials, the number of locations and seasons, and the choice of the ‘comparator’ have considerably differed from case to case. It is hardly plausible that the compositional data have been analysed in a statistically sound way.
• The requests for authorisation of Bt11 sweet maize and GA21 have been submitted to the Dutch authorities, which treated the data submitted in a rather critical way. Applicants were urged to provide additional data on the content of five secondary metabolites in the GM maize plants compared to their non-GM controls to underpin the degree of substantial equivalence. Such data were provided, but it did not lead unambiguously to a determination of (the degree of) substantial equivalence of GA21. The Dutch authorities noted that applicants would be helped by concrete guidance concerning the number of samples, locations and years, which would be needed for the quantitative analyses.
• In both these cases of a ‘heavy’ authorisation procedure, several member states raised critical questions on the assessment by the Dutch authorities. In the case of Bt11 sweet maize the applicant has provided a response to these questions. In the case of GA21 it is not clear whether the applicant has provided additional information. In both cases it is unclear how these questions and responses impact regulatory decision-making by the European Commission and national authorities of EU member states, as the EC Scientific Committee on Plants has determined Bt11 (sweet) maize as well as GA21 maize as substantially equivalent to their non-GM controls.
• The request for authorisation of crosses of T25 and MON810 has been submitted to the Dutch authorities, which have not yet completed their initial assessment. The EC Scientific Committee on Plants concluded that T25xMON810 hybrids are substantially equivalent to T25 and MON810 and non-GM maize.

The UK authority received the original submission in August 1994. In February 1995 the UK authority approved the food use of paste of two GM tomato hybrid lines. In August 1995 the applicant requested to extend the scope of the original approval to tomato paste and processed tomato products containing the paste, like ketchup and pizza sauces. The request was also forwarded to extend the approval to any tomato line containing the construct pJR16S. The UK authority agreed with extending food safety clearance to paste and products containing the paste but not to any line containing the construct.

Comparative nutritional and toxicological analyses were carried out on GM tomatoes and commercially available, conventional tomato varieties produced during trials in one year. Nutritional values fell within the range of controls and the data on four known inherent tomato toxins (alpha-tomatine, solanine, chaconine and nicotine) indicated that their levels in both the GM and non-GM samples were below the level of detection. The UK authority concluded that the nutritional and toxicological composition of the GM tomato was comparable to that of conventional tomatoes.

In 1997 Regulation 258/97 came into force. The applicant submitted a request for a safety evaluation of TGT7F by the UK authority in March 1998. Data were provided on the content of moisture, ash, fat, total carbohydrate, total dietary fibre, soluble fibre, insoluble fibre, total sugars, protein, cellulose, hemicellulose, energy value, Na, K, Ca, Mg, P, Fe, vitamins A, E, B1, B2, B6, C, niacin, folic acid, malic acid, citric acid and lycopene. On this basis the UK authority concluded that the genetic modification had not changed the nutritional profile. Further, other inherent tomato toxins, such as tyramine, serotonine, histamine, nicotine and lectins, were also analysed. Their levels in GM tomato paste were comparable to those in the controls. In an exemplary case study on the application of the concept of substantial equivalence by the OECD in 1993, it was suggested to include inherent tomato toxins, like alpha-tomatine, tomatidine, aglycone of tomatine, saponines, coumarins, lectins, serotine, oxalate, protease-inhibitor and histamine

Levels of toxic metals (As, Cd, Pb and Hg) in the GM tomato fell within the range of values obtained for its control. It was further concluded that the NPTII gene encoding for kanamycin-resistance and its protein, which could not be detected in processed product from the GM tomato, were not inherently toxic. According to the UK authority, no nutritional and toxicological differences existed between the GM tomato and its non-GM counterpart. This was seen as further evidence that no secondary effects had occurred as a result of the genetic modification.

In June 1998 the EC Scientific Committee on Plants concluded that the harvested GM tomato fruit is substantially equivalent to fruit of other commercial processing tomatoes.

In September 1999 the EC Scientific Committee on Food confirmed the opinion of the SCP that the GM tomato is substantially equivalent to unmodified and conventional counterparts except for the introduced traits.

• The applicant sought a (full) safety evaluation of processed products of GM tomato TGT7F. The UK authority concluded that no nutritional and toxicological differences existed between the GM tomato and its control. The UK authority did not explicitly establish the GM tomato as substantially equivalent, whereas the EC Scientific Committee on Plants and the EC Scientific Committee on Food both concluded that the GM tomato is substantially equivalent.
• Compositional data have been obtained from trials during one year. It is not clear whether these data have been analysed in a statistically sound way.
• Comparative data of the GM tomato and its control regarding several inherent tomato toxins, such as tomatidine, aglycone of tomatine, saponines, coumarins, protease-inhibitor and oxalate, were not provided.

In the early 1970s the first so-called "recombinant DNA" (rDNA) techniques were developed for splicing DNA molecules and reintroducing them into living organisms. Some molecular biologists immediately expressed concern about specific dangers from the new tools, which might allow production of harmful viruses on a much larger scale than would otherwise be possible, or to transfer genes for antibiotic resistance into disease-causing bacteria, thereby rendering antibiotic therapy ineffective. This prompted the US National Academy of Science to convene a committee that was chaired by Paul Berg. This committee distributed a letter, (sometimes referred to as the "Berg" letter) calling for a voluntary moratorium on the cloning of genes for antibiotic resistance genes and of DNA from animal viruses pending an international conference to discuss safety issues related to these rDNA techniques. That conference¹ was held at Asilomar, California, in 1975, which set in motion the first discussions on how to regulate the safety of recombinant DNA research in the US and later in almost every industrialised country.

A couple of years later rDNA techniques were used to develop novel organisms for experiments and usage outside laboratories. In 1983 this led the Committee for Scientific and Technology Policy of the Organisation of Economic Co-operation and Development (OECD) to create a Group of National Experts on Safety in Biotechnology (GNE). In 1986 the work of the GNE resulted in general safety guidelines² for the use of genetically modified organisms (GMOs) in industry, agriculture and the environment. These guidelines recommended focusing safety concerns on whether organisms modified by recombinant DNA techniques pose an ‘incremental’ risk to organisms modified by conventional means. While rDNA techniques may result in the production of organisms expressing a combination of traits that are not observed in nature, genetic changes from rDNA techniques will often have inherently greater predictability compared to traditional techniques, because of the greater precision that rDNA technique affords to particular modification. On this conceptual basis³ the GNE expected that any risk associated with the application of rDNA modified organisms may be assessed in generally the same way as those associated with non-rDNA organisms.

Four years later, in 1990, the GNE agreed to establish a Working Group for developing scientific principles that would focus on the safe use of new food or food components derived by modern biotechnology. In 1993 OECD⁴ published the results, which indicated that the "scientific approach to safety evaluations of new foods or food components derived by modern biotechnology (…) is based on a comparison with traditional foods that have a safe history of use. This approach is based in turn on the concept of substantial equivalence, which articulated procedures used in the past, albeit intuitively, for accepting new foods". According to the report, the safety of food for human consumption is based on the concept that there should be a reasonable certainty that no harm will result from intended uses under the anticipated conditions of consumption. Historically, foods prepared and used in traditional ways have been considered safe on the basis of long experience even tough they may have contained natural toxicants or anti-nutritional components. According to the working group, modern biotechnology broadens the scope of the genetic changes that can be made in food organisms. This does not inherently lead to foods that are less safe than their conventional counterparts and therefore safety evaluations of foods derived by modern biotechnology do not necessitate a fundamental change in established principles, nor do they require a different standard of safety. Moreover, the precision inherent in the use of certain molecular techniques should enable direct and focussed safety assessment where such assessment is desired. In the view of the Working Group, the most practical approach to the determination of safety of foods derived by modern biotechnology is to consider whether they are substantially equivalent to analogous conventional food product(s), if such exists. A demonstration of substantial equivalence takes into consideration a number of factors, such as:

• Knowledge of the composition and characteristics of the traditional or parental product or organism.
• Knowledge of the characteristics of the new component(s) or trait(s) derived, as appropriate, from information concerning: the component(s) or trait(s) as expressed in the precursor(s) or parental organisms; transformation techniques (as related to understanding the characteristics of the product) including the vector(s) used and any marker genes used; possible secondary effects of the modification; and the characterisation of the component(s) or trait(s) as expressed in the new organism.
• Knowledge of the new product/organism with the new component(s) or trait(s), including the characteristics and composition [i.e. the amount of the components or the range(s) of the expression(s) of the new trait(s)] as compared with the conventional counterpart(s) (i.e. the existing food or food component).

In addition, the principle was set out that if a modified food or food component is determined to be substantially equivalent to an existing food, it can be treated in the same manner. No additional safety concerns would be expected. However, when a product is determined not to be substantially equivalent, the identified differences should be focus of further evaluations. In case there is no basis for comparisons, that is, where no counterpart has been previously consumed as food, the new food or food component should be evaluated on the basis of its own composition and properties.⁵

In 1996, three years after the OECD had introduced the concept of substantial equivalence an expert consultation was convened jointly by the Food and Agriculture Organisation (FAO) and World Health Organisation (WHO). This Consultation⁶ recommended - in line with OECD - that substantial equivalence is an important component in the safety assessment of foods and food ingredients derived from genetically modified plants intended for human consumption.

On October 7^th, 1999, the authoritative scientific journal Nature⁷ published a critical view on the concept of substantial equivalence. In the authors’ view, the degree of difference between a natural food and its GM alternative before its ‘substance’ ceases to be acceptably ‘equivalent’ had not been not defined, nor had an exact definition been agreed by legislators. The vagueness of this concept might make the concept useful to industry but unacceptable to the consumer. The authors therefore argued that legislators should have treated GM foods in the same way as novel chemical compounds such as pharmaceuticals, pesticides and food additives. Companies should have been required to conduct a range of toxicological tests to set ‘acceptable daily intakes’ (ADIs). Although the authors acknowledged that toxicological and biochemical tests, and their interpretation, are notoriously problematic and contested, and are slow and expensive, they could provide information vital to consumer protection.

Subsequently, criticism was expressed that the concept of substantial equivalence had been misapplied in the case of genetically modified, glyphosate-tolerant soybeans⁸, even on its own terms. With reference to a report⁹ by a Dutch government team, the authors reached the final conclusion that "substantial equivalence is inherently anti-scientific because it was created primarily to provide an excuse for not requiring biochemical and toxicological tests." They ended by recommending "to replace the concept with a practical approach which would actively investigate the safety and toxicity of GM foods rather than merely taking them for granted".

The OECD responded^10,11 that the substantial equivalence was not developed as an excuse for not requiring biochemical or toxicological tests but emerged from a number of observations. The first and most significant of these was that new varieties of whole foods coming to the market do not undergo extensive toxicological testing. The second observation was the knowledge that traditional whole foods often contain natural toxicants and/or anti-nutritional substances, so that it would be likely that a battery of traditional toxicological tests might well reveal undesirable effects in animal tests. Despite this, whole foods if prepared and used in appropriate ways have been considered safe on the basis of long term experience.

The OECD further argued that in assessing the safety of a GM food, it would be unreasonable and unscientific to expect an absolute standard of safety, which could be demonstrated through traditional testing. Therefore a notion was required that a new modified variety should be as safe as traditional varieties. So the important starting point in the safety assessment should be a comparison between the new genetically modified variety and its traditional counterpart, that is a determination of substantial equivalence. OECD emphasised that substantial equivalence is not in itself a substitute for a safety assessment. It is a guiding principle, which is intended to be a useful tool for those engaged in safety assessments. Finally, the OECD pointed out that there is work under way to identify the critical nutrients and toxicants found in major crop plants, which would be an important focus of a demonstration of substantial equivalence.

In his report¹² the hairman of the OECD Edinburgh Conference on 28 February and 1 March 2000 concluded that safety assessment of any novel food involves a variety of kinds of evidence. One commonly used tool was the concept of substantial equivalence. The essence of this idea was that a comparison between the novel food and one already in the diet provides the basis for asking questions about the safety of the novel products. Substantial equivalence was not a quantitative criterion or hurdle. According to the report, it was continually and updated, but it was timely now, after six years of using the tool, to undertake a more detailed review.

In June/July 1999 the Codex ad hoc Intergovernmental Tasks Force on Foods Derived from Biotechnology had been established by the twenty-third session of the FAO/WHO Codex Alimentarius Commission. Nearly a year later, in March 2000, in its first session, this Task Force¹³ welcomed the initiative of FAO and WHO to convene another joint expert consultation two months later, in May/June. The Task Force also identified, among others, the following questions for this joint expert consultation: What is the role, and what are the limitations, of substantial equivalence in the safety and nutritional assessment? And are there alternative strategies to substantial equivalence that should be used for the safety and nutritional assessment?

In May 2000 the OECD Task Force for the Safety of Novel Foods and Feeds reported¹⁴ that in 1996 an FAO/WHO consultation had endorsed the application of substantial equivalence. It had recognised that establishment of substantial equivalence is not a safety assessment per se, nor intended to be a measure of absolute safety. Establishment of the new GM food as substantially equivalent implies that it will be no less safe than the conventional food, while demonstrating absolute safety is an impractical goal. A key criticism had been that substantial equivalence is not measurable and therefore inappropriate to safety assessment. However, in the view of the Task Force, the concept of substantial equivalence is a tool, which helps to identify any difference, intended or unintended, that might be focus of further safety evaluation, involving traditional nutritional, toxicological or immunological testing, or long-term studies, depending on the identified differences. Determining substantial equivalence entails consideration of:

• The trait encoded by the genetic modification.
• Phenotypic characterisation of the new food source, compared with an appropriate comparator already in the food supply.
• Compositional analysis of the new food source or the specific food products compared with the selected comparator.

Because it is a comparative process for safety evaluation, the determination of substantial equivalence can be performed at several points along the food chain (e.g. at the level of the harvested or unprocessed food product, individual processed fractions or the final food product or ingredient). Although from a practical point (e.g. where multiple fractions from a single source will be used as different food products) substantial equivalence should be typically determined at the level of the unprocessed food product, the flexibility of the concept permits the determination to be targeted at the most appropriate level, based upon the nature of the product.

The Task Force further discussed why the choice of an appropriate comparator having an acceptable history of safe food use is crucial to effective application of the concept of substantial equivalence of a new GM food. Sufficient analytical data should be available in literature, or be generated through analysis, to allow a comparison to be made. Guidance on applying the concept had been developed to provide clarity in interpreting the criteria that would constitute an appropriate framework for determining substantial equivalence. The Task Force also pointed out that the nature of the comparative approach with respect to nutrients limits its universality, since the relevance of nutrients in a particular crop depends on consumption patterns that may vary from region to region. Therefore, where such differences exist, these must also be considered in identifying the key nutrients to be assessed.

The FAO/WHO joint expert consultation¹⁵ in 2000 noted that since 1996 several countries had used the concept of substantial equivalence and had found this approach to be scientifically sound and practical. However there had not been universal consensus on the application of this concept. This had resulted in criticism that substantial equivalence did not provide a sufficient basis for safety and in calls from national governments and international bodies to consider alternative approaches. In 2000 FAO and WHO therefore convened another expert consultation to review the scientific basis, application, and limitations of the concept of substantial equivalence.

One of the papers¹⁶ discussed by the Consultation summarised differences between chemical and food toxicity evaluation in the following table.

The Consultation argued that in contrast to many compounds such as pesticides, pharmaceuticals, industrial chemicals and food additives, whole foods are complex mixtures of compounds characterised by wide variation in composition and nutritional value. Practical difficulties of obtaining meaningful information from conventional toxicology studies on the safety of whole foods, which had become particularly apparent when animal feeding studies were used to assess the safety of irradiated foods, meant that an alternative approach was required for the safety assessment of genetically modified foods. This has led to the development of the concept of substantial equivalence. The Consultation agreed that the practical difficulties in the application of conventional toxicology studies to whole foods preclude their use as a routine safety assessment technique for genetically modified foods. Further, from an animal welfare perspective the Consultation recognised that the use of toxicology studies could not be justified where it was unlikely to result in meaningful information. It also noted that the concept of substantial equivalence was a good example of an approach to reduce the use of animals in toxicology studies by refining safety assessment techniques and replacing animal models with alternatives.

The paper¹⁷ ‘Beyond substantial equivalence’ published by Nature in October 1999 was also presented at the meeting. The Consultation acknowledged that the concept of substantial equivalence had attracted criticism. In part, this related to the mistaken perception that determination of substantial equivalence was the end point of safety assessment rather than a starting point. It is a concept used to identify similarities and differences between the genetically modified food and its traditional counterpart. Intended and unintended differences identified then become focus of the safety assessment. The potential occurrence of unintended effects is not specific to the use of recombinant DNA techniques, in the view of the Consultation. It is an inherent and general phenomenon that can occur in conventional breeding, in which one of the approaches to cope with this problem is to select and discard plants with unusual and undesired phenotypic and agronomic parameters at an early stage of plant variety development. Consecutive back-crossing is also a common procedure to eliminate unintended effects. Unintended effects due to genetic modification may be ‘predictable’ based on metabolic connections to the intended effect or knowledge of the site of insertion, or they may be ‘unexpected’.

The Consultation further indicated that the comparator used to detect unintended effects should ideally be the near isogenic parental lines grown under identical conditions. In practice, this may not be feasible at all times, in which case a line as close as possible should be assessed. The resulting natural variation should be taken into account in assessing the statistical significance of the unintended effect. Where statistically significant unintended effects are observed, their biological significance and safety should be assessed. For that purpose, data on the genetically modified plant should be compared to data on other conventional varieties and literature data. If the differences exceed natural variations in traditional food crops, further assessment is required. Data for comparison should be obtained using validated methods and analysed using appropriate statistical techniques.

One of the papers¹⁸ discussed by the Consultation indicated that applicants for marketing of food produced from GMOs had normally used the comparison principle of substantial equivalence but there had been a lack of consistency from case to case in the data provided, even within the same plant species. This inconsistency and sometimes lack of data demonstrated the need for guidelines in order to harmonise the basis data requirements for comparison. The paper recommended therefore developing consensus documents and/or databases for different species or organisms and foods and feeds. Guidelines should also be developed for key substances, nutrients and data needed for assessment of substantial equivalence including for each substance the variation related to the history of the food and feed and its intake. In addition, guidelines should be developed for appropriate design of field trial with GM food crops and for appropriate statistical methods.

The Consultation further indicated that present approaches to assess possible unintended effects are based, in part, on the analysis of specific components (targeted approach). In order to increase the probability of detecting unintended effects, profiling techniques are used at different level, e.g. genomics, proteomics and metabolomics, and may contribute to the detection of differences in a more extensive way than targeted chemical analysis. However, they are not yet fully developed and validated and have certain limitations.

Finally, the Consultation was also of the view that there were no alternative strategies that would provide a better assurance of safety of genetically modified foods than the appropriate use of the concept of substantial equivalence. Some steps in the food safety assessment could nevertheless be refined by further developing and validating the new profiling techniques, as these would provide the means of a more detailed comparison.

In the fall of 1999 the US National Research Council published a first Internet version of a report¹⁹ by an expert committee that had discussed the science and regulation of genetically modified plants. The expert committee first task was to review the three principles formulated by the National Academy of Science (NAS) in 1987:

1. There is no evidence that unique hazards exist either in the use of rDNA techniques or in the movement of genes between unrelated organisms.
2. The risks associated with the introduction of rDNA-engineered organisms are the same in kind as those associated with the introduction of unmodified organisms and organisms modified by other methods.
3. Assessment of the risks of introducing rDNA-engineered organisms into the environment should be based on the nature of the organism and the environment into which it is introduced, not on the method by which it was produced.

The committee pointed out that it was not aware of controlled field studies, which directly compare the ecological effects of transgenic and conventional pest-protected plants bred for the same pesticidal trait. The committee indicated that its conclusions about these principles were therefore not based on data from such comparisons, but on mechanistic knowledge and scientific information about the resulting GM plants. It further argued that both conventional breeding and genetic modification have the potential to produce organisms of high or low risk, and therefore the properties of the organism should be the focus of risk assessment, and not the process by which it is produced.²⁰ With respect to the potential that health impacts of the consumption of GM plants may occur due to pleiotropic effects of genetic modifications, the committee concluded that monitoring of pleiotropic changes in plant physiology and biochemistry should be an important element of health safety reviews. Therefore it recommended research to assess and enhance data on the baseline concentrations of plant compounds of potential dietary or other toxicological concern, and to determine how concentrations of these compounds may vary depending on the genetic background of the plant and environmental conditions. It also recommended that the three regulatory agencies Environmental Protection Agency (EPA), Food and Drug Administration (FDA)^21,22 and United States Department of Agriculture (USDA) collaborate on the establishment of a database for natural plant compounds of potential dietary or other toxicological concern.

In February 2001, under the heading ‘Elements of Precaution’, an expert panel convened by the Royal Society of Canada²³ published recommendations for the regulation of food biotechnology in Canada. In relation to the concept of substantial equivalence, the expert panel noted that conceptual and practical²⁴ implementation of this concept is the most critical element in current approval process. In the view of this expert committee the concept was clearly rooted in the existing paradigm for new crop development through traditional methodologies. A breeder who has genetically manipulated a crop through crossing/selection takes it as a given that, despite the numerous small changes introduced into the genome of the new genotype, the species as an entity remains largely unmodified. The expert panel however argued that present framing of the concept of substantial equivalence links it intimately with the definition of ‘novel trait’ in a way that leads to a logical impasse. This logical confusion has led to an ambiguous use of ‘substantial equivalence’ in the regulatory world, To the opinion of the expert committee, there were now two different kinds of the use of this concept: the decision threshold interpretation and the safety standard interpretation. The expert committee recommended the latter. "A GM organism is substantial equivalent if rigorous scientific analysis establishes that, despite all changes introduced into the organism as a result of the introduction of novel genes, the organism poses no more risk to health or to the environment than does its conventional counterpart."

Under the heading "Can substantial equivalence become scientifically rigorous?" the expert panel therefore suggested an integrated approach at six relevant levels (genome, transcript, protein, metabolite, health impacts, environmental impacts). The answers obtained from the molecular analyses, in particular at levels 1 to 4, would speak directly to the validity of the simple linear model of ‘precise’ genetic engineering. If these analyses are conducted on a range of existing transgenic varieties and the predictions of the simple linear model prove to be valid, that outcome would provide essential scientific support for the current regulatory view that the insertion of the transgene(s) has created no significant changes in the original variety other than those predicted and desired. If, on the other hand, the molecular analyses demonstrate that the simple model is not valid, the data would provide immediate entry points for studying the impacts of the detected changes on human health and the environment. The outcome of those follow-up studies will then help determine whether the impacts create a significant risk.

The European Community Regulation²⁵ (EC) 258/97 on Novel Foods and Food ingredients sets out rules for notification or authorisation and labelling of GMO-derived food products and other novel foods.

In general²⁶ the authorisation for food use of a GMO-derived product is a one-step process if all Members States agree to the initial assessment of a Member State, and a two-step process if one or more Member States object. The first step is an initial assessment by the Member State where the food is to be placed on the market first. In case of a favourable opinion, this Member States informs the other Member States via the Commission. If there are no objections against the application, the first Member State can authorise for the whole Community and the product can circulate freely in the Single Market. This procedure can be completed within several months. If there are objections by Member states, the second step has to be taken, which means a Commission Decision, following consultation of the Scientific Committees, such as the Scientific Committee in Foods (SCF) and the Scientific Committee on Plants (SCP) on matters relating to public and the Comitology procedure. Once at community level, the time frame for authorisation is necessarily extended.

As a derogation²⁷ from the full authorisation procedure, the Novel Foods Regulation provides for a simplified procedure for foods derived from GMOs but no longer containing GMOs which are "substantially equivalent" to existing foods with respect to composition, nutritional value, metabolism, intended use and the level of undesirable substances. In such cases the companies only have to notify the Commission when placing a product on the market together with either scientific justification that the product is substantially equivalent or an opinion to the same effect delivered by the competent authorities of a Member State. The product can then be marketed in the entire EU. It should however be noted that in all cases of notifications analysed in this report, the SCP and/or SCF have also issued an opinion on whether the (ingredients of) GM food crops were to be viewed as substantial equivalent (or not). Both scientific committees fall under the European Commission Directorate General Health and Consumer Protection.

Noteworthy is the role of SCP as it has also delivered opinions on the risk assessment of these GM food crops as required by Directive 90/200 regulating experimental and commercial releases into the environment of GM plants, for which the European Commission Directorate General Environment is the main responsible body.

The following table shows examples of GM crops or ingredients derived thereof, which have been notified (or authorised) for food use in the European Union until May 2001. In practice mainly the UK, Germany and The Netherlands have received requests for notification or authorisation of GM food crops.

• ACNFP: Advisory Committee on Novel Foods and Processes (United Kingdom); BgVV: Bundesamt für gesundheitliche Verbraucherschutz und Veternärmedizin (Germany); VNV: Veiligheidscommissie Nieuwe Voedingsmiddelen (The Netherlands).

Paragraph 1 of Article 14 of Regulation 258/97stipulates that no later than five years from the date of entry into force and in the light of the experience gained, the Commission shall forward to the European Parliament and to the Council a report on its implementation. The date of entry was 27 January 1997. This implies that the Commission should forward this report at the latest on 27 January 2002.

As part of the development of Regulation 258/97 the European Commission asked the Scientific Committee for Food (SCF) to develop recommendations²⁸ concerning the scientific aspects necessary to support an application on the market of novel foods and novel food ingredients. With reference to OECD and WHO the SCF reminded that establishment of substantial equivalence is not a safety or nutritional assessment in itself but an approach to compare a novel food with its conventional counterpart.

Substantial equivalence may be established either for the whole food or food component including the introduced ‘new’ change, or it might be established for the food or food component except for the ‘new’ change introduced. If substantial equivalence of a novel food or food component has been established, it means that their safety is comparable to that of their conventional counterparts. If a novel food has not been found to be substantially equivalent, this does not imply that it is unsafe. It just indicates that it should be evaluated on the basis of its unique composition and properties.

In the view of the SCF, analytical studies are of crucial importance not only for the establishment of substantial equivalence but also as a prerequisite for nutritional and toxicological assessments. Methods applied have to be standardised and validated to ensure quality and consistency of data. Investigations should especially focus of the content of critical macro- and micro-nutrients, any critical toxicants and anti-nutritional factors, which might be either inherently present or due to the modification.

On 18 December 1998 the Scientific Committee on Plants (SCP) of the European Commission published a guidance document²⁹ to facilitate applicants in the preparation of GM plant dossiers.

With respect to the establishment of substantial equivalence, SCP recommended that data should be obtained from a valid comparison of GM plants and a conventional line, which is preferably isogenic, based on samples from at least two seasons, grown at a number and variety of geographical locations. The data should further be accompanied by an appropriate statistical treatment. The specific analyses would depend on the plant species examined. Applicants were also encouraged to provide data, which demonstrate the range in component concentrations, found in non-GM counterparts and thereby to make comparisons with the GM plant in question.

In 1999 the UK competent authority, the Food Standards Agency, presented to the European Commission a paper³⁰ to promote discussion about the provision of statistically valid data in applications for products approvals under the EC Novel Foods Regulation. The paper resulted from work by the UK Advisory Committee on Novel Foods and Process (ACNFP).

As starting point this paper took the 1996 FAO/WHO expert consultation statement that "substantial equivalence is established by a demonstration that the characteristics assessed for the GMO, or the specific food product thereof, are equivalent to the same characteristics of the conventional comparator. The levels and variation of characteristics (e.g. phenotypic and composition characteristics etc.) in the GMO must be within the natural range of variation for those characteristics considered in the comparator and be based upon an appropriate analysis of data".

The paper then detailed the factors, which might affect ‘natural variation’, the confidence limits and sample size, and the number of trial sites, years and replicates. It concluded that statistically valid data are necessary for effective comparison of a GM crop with a non-GM control. The use of such data would minimise the effects of natural variation and would indicate whether there are any differences between the GM crop and the non-GM crop, which require further investigation. Significant difference between the results of proximate and detailed chemical analyses of a GM crop product and its non-GM control may be an indication of secondary or pleiotropic effects. It further recommended that trials with GM crops be conducted under the range of environmental conditions in which it is intended that the GM crop should be grown commercially. A minimum of six sites should be used, with the number of replicates to be determined for individual crop species, and trials should be carried out for a minimum of two successive years. The paper further recommended that the Commission seek EU-wide agreement on an appropriate confidence interval to use in the statistical analyses and the sample size. The paper noted that 95% confidence intervals are commonly used. In addition, data from different sites should not be pooled before being statistically evaluated.

With reference to the FAO/WHO Consultation of May 2000, the European Commission Scientific Steering Committee³¹ (SSC) was due to the opinion that the establishment of substantial equivalence should be used as a starting point to guide the safety evaluation. It noted that an operational definition was needed as the various European Commission scientific committees had applied the concept on a case by case basis. It proposed that this issue should be the focus of a future opinion.

In the view of the SSC, the routine analytical procedures in use still needed to be standardised in terms of sampling and extraction procedures, validation of profiling methods and in bio-informatics. Especially if transgenic material is introduced by gene technology³² or by modern conventional plant breeding to alter an existing metabolic pathway or introduce a new one, toxicants might accumulate. The methods applied did not provide certainty to detect all new non-targeted toxic plant constituents or an increase of the amount of unidentified existing toxic constituents. The methods therefore, needed to be expanded. The SSC expected that tools (under development)³³ such as genomics, proteomics and metabolomics would offer greater opportunities for their food safety evaluation.

In a review article³⁴ on the potential pleiotropic effects of genetic modifications on constituents of modified crops data from US and EC documents were investigated with regard to inherent plant toxins and anti-nutrients: rape (glucosinolates, phytate), maize (phytate), tomato (tomatine, solanine, chaconine, lectins, oxalate), potato (solanine, chaconine, protease inhibitors, phenols) and soybean (protease-inhibitors, lectins, isoflavones, phytate). According to the authors, genetic modification of food-plants is developing rapidly, but until now no long-term experience on nutritional (or ecological) effects is available. In the view of the authors, especially the relevance of pleiotropic effects due to the genetic modification is unclear, but could play a role in the question of possible changes in the expression of constituents such as inherent plant toxins and anti-nutrients. The impossibility of a prediction of the integration region of the genetic modification is one of the main problems for the estimation of the probability and of the dimension of pleiotropic effects. The authors recognised that the concept of substantial equivalence had been seriously criticised (Millstone et al., 1999)³⁵ but strong scientific support had defended the use of this concept. Although an assessment of substantial equivalence is necessary for notification in the EU, the authors’ analysis of selected notification documents of GM plants indicated that relevant data about inherent plant toxins and anti-nutrients were often missing. Some documents provided no data at all about inherent plant toxins and anti-nutrients, and in a few cases applicants argued that analysis of some constituents was not relevant. In fact, there were no coherent regulations to which companies could adhere for the selection of inherent plant toxins and anti-nutrients that have to be analysed. The documents showed, in the view of the authors, no consistency. The assessments of substantial equivalence could therefore hardly be conclusive. Levels of inherent plant toxins and anti-nutrients were found within the range of the non-transformed parental plants in most GM plants. However, in several documents no or incomplete data on these compounds could be found, for example with regard to saponins in tomato (Zeneca, 1996) or phenolic compounds and couramins in potato (Avebe, 1996). In the dossier of GM tomato many anti-nutrients were tested, such as glycoalkaloids and lectins, but no data were provided about oxalate (Zeneca, 1996). In addition, only in one of the four notification of GM maize varieties phytate levels were analysed. Since, according to the authors, phytate occurs in considerable amounts in maize, its levels should be analysed for an assessment of the substantial equivalence of the GM plant. The silage and grain of glufosinate-tolerant maize (AgrEvo, 1995) was found not to be different from commercial varieties in essential nutrients or inherent plant toxins and anti-nutrients. In preliminary documents of the GM glyphosate-tolerant maize line GA21 (Monsanto 1998), compositional components, such as protein, fat, ash, carbohydrates, moisture and fibre, amino acid composition and fatty acid profile, calcium and phosphorus were analysed, but no inherent plant toxins and anti-nutrients, such as phytate, were included in the analyses.

Further, in one dossier on modified rape plants (Plants Genetic Systems, 1996) the values of glucosinolates in the modified plant were significantly different (higher) than in the non-transformed plants, and in some cases significant differences between the glucosinolate content of one modified line and its non-transgenic counterpart were observed. It was noticed that the glucosinolate content varied more between the different locations than between the different transgenic and non-transgenic entries in a given location. These variations were discussed by the applicant, suggesting it is generally accepted that commercial food derived of plants exhibits considerable variability in its compositions and that this variability is more the result of the interaction of the genotype with the environment, rather than the results of the insertion of specific genes into the plant genome. But, in the view of the authors, this conjecture may not be scientifically justified: how much is the accepted range of the variability, what are the regulations for variations above the statistical limits, what variations may be dangerous? These questions were not completely clarified yet. The amounts and natural variation of anti-nutrients in one crop plant species could differ considerably due to factors such as state of ripening, year of production, storage, varietal differences, growing conditions (climate, soil) and stress or pathogen infection. Literature data sometime showed very wide variations in typical plant constituents. And although ranges for most of the compositional variables were available in literature, these data might not be directly comparable due to differences in analytical methods or sample preparation. In addition, much of the literature data were relatively old and might not encompass the compositional variables of modern crop varieties. Comparisons between studies were often further complicated because parameters of analysis like units and basis of standardisation of substances showed no conformity and were difficult to interpret.

In conclusion, the authors argued that their review supported the usefulness of the concept of substantial equivalence. But they viewed the lack of specification as to which key components have to be analysed in a certain GM plant to establish substantial equivalence as a major problem. A minimum list³⁶ of macro- and micro-nutrients, secondary plant constituents, inherent plant toxins and anti-nutrients and allergens known to be associated with a crop species that should be analysed in order to determine substantial equivalence should be established. Another difficult problem for determining substantial equivalence was that it is not always technically possible to use authentic isogenic control lines. The process to obtain all the required permissions takes a long time, and in the end the compared data from transgenic lines often come from very early transformants and are not really consistent with the final products. In addition, special care has to be taken in investigating and controlling possible effects of environmental conditions on constituents of GM crops. Although such effects, possibly caused by unusual environmental parameters, could similarly be seen with conventional crops, several problems to agronomic characteristics have been observed³⁷ in GM crops due to unusual temperatures and possible changes in plant physiology caused by the insertion of genes. In the view of the authors, such observations, together with a specific public awareness, suggest that conclusive information requirements for notifications and a scientifically reviewed, publicly available risk assessments, as well as post-marketing controls would be important to safely use gene technology in food and feed production.

One of the key issues in the safety assessment of GM food crops is to determine whether ‘unintended’ effects may have occurred as a result of the genetic modification. However, in the view of Noteborn³⁸ (2000), a compositional analysis based on single compounds has its limitations with respect to (unknown) anti-nutrients and natural toxicants. Novel technologies and detailed analyses of the (food) foreign gene products and of possible ‘unintended’ effects in the plant were therefore urgently needed. In the study the utility of a combination of nuclear magnetic resonance (NMR) spectroscopy and liquid chromatography (LC) for the multi-component comparison of low-molecular compounds (i.e. chemical fingerprinting) in complex plant matrices was demonstrated and discussed. This methodology might further contribute to the demonstration of substantial equivalence by its ability to compare compositional changes with regard to a couple of hundred low-molecular compounds between a GM food crop and related non-GM reference lines. The study used two different kinds of GM tomatoes as model crops: a GM tomato with a down-regulated expression of exogalactanase leading to altered texture, mouthfeel and colour and a GM tomato expressing a Bacillus thuringiensis (cry1Ab5) gene conveying resistance to Lepidoptera pests.

The study also resulted in a proposal for a hierarchical approach by comparing the chemical fingerprints of the GM plant to those of: 1) isogenic parental or closely related lines bred at identical and multiple sites; 2) extended ranges of commercial varieties of that plant; and 3) downstream processing effects. In the view of the authors, this would be important to assess the likelihood that the statistical differences in a GM crop plant might be false positives due to chance alone, or arose from natural genetic and/or physiological variations.

In April 2001, under the remit of the Scientific Committee on Foods (SCF), the Scientific Committee on Animal Nutrition (SCAN), and the Scientific Committee on Plants (SCP)³⁹, a SCP working group of experts started to draft a guidance document for applicants, who are seeking authorisation for a GMO release and subsequently authorisation for their food anduse. This guidance document should in principle be adopted by the Scientific Steering Committee (SSC). It would provide guidance for the molecular characterisation of the gene construct, the establishment of substantial equivalence, performing the safety assessment, the nutritional assessment and the environmental assessment.

1. Application file C/DE/98/6.
2. Refined oil derived of Liberator pHoe6/Ac notified by Germany on 21 October 1999. Correspondence between BgVV, the European Commission Enterprise Directorate-General and the Dutch Committee Safety Evaluation of Novel Foods and personal communication with Mrs Schauzu of BgVV on May 3^rd, 2001.
3. Additional information from Aventis on 19 June 2000.
4. Opinion of Scientific Committee on Plants issued on 30 November 2000.

The applicant submitted data (minimum and maximum values) on the contents of oil, ash, crude fibre, sinapine, phytate, glucosinolates and amino acids in desolventising meal of GM oilseed rape plants, control-plants and commercial varieties. Data on levels of chlorophyll, tocopherols, sterols, P, Fe, Cu, Ca, Mg, S, erucic and fatty acids in blending crude oil and deodorised oil were also provided. Samples were pooled from the harvest of two different locations in Canada in 1997. Although the data were not statistically analysed, the applicant concluded that no changes in most of the important compounds in the GM oilseed rape plants compared to its controls had occurred.

BgVV’s initial assessment is based on semi-quantitative data on the contents of protein, oil and glucosinolates and on qualitative and quantitative data on amino acid composition of the flour, fatty acid composition of the seed and oil, of which also the contents of sterols and tocopherols have been analysed. Protein content of refined oil of Liberator pHoe6/Ac has been determined at 0.01 mg/ml based on data submitted in the case other GM oilseed rape lines, i.e. Topas 19/2 and MS8xRF3. According to BgVV, refined oil of Liberator pHoe6/Ac is substantial equivalent to that of its comparator, the parental line Liberator regarding nutritional value, intended use, composition and the level of undesirable substances. On 21 October 1999 the applicant notified the refined oil of Liberator pHoe6/Ac to the European Commission, which subsequently informed the other member states about this notification on 8 November 1999.

The correspondence between the Commission and the member states shows that a few member states noted that the data were obtained from winter oilseed rape varieties only, while consent for commercial release of transformation event Liberator pHoe6/Ac might also include spring oilseed rape varieties based on this transformation event. Further, Denmark did not view this GM oilseed rape plant as substantial equivalent. It noticed from the data submitted that in one of two lots the GM oilseed rape plants had a 50% higher erucic acid content than their unmodified counterparts, whereas in the other lot the GM oilseed rape plants showed a 35% decrease. Levels of tocopherols were 20 to 45% higher in the GM oilseed rape plants than in their counterparts. Chlorophyll content in the GM oilseed rape plants increased with 400% (which also led to questions by The Netherlands), and desolventising meal of the GM oilseed rape plants contained 20 to 25% more glucosinolates.

In response to questions raised by Italy, Portugal, United Kingdom and Ireland, Aventis submitted additional information confirming that no modified DNA or modified protein is present in refined oil. Aventis further provided technical information about the application of High Performance Liquid Chromatography (HPLC) to determine the content of glucosinolates in the oil of MS8xRF3, respectively Gas Liquid Chromatography (GLC) in the case of Falcon GS40/90 and Near Infrared Reflectance Spectroscopy (NIRS) in the case of Liberator pHoe6/Ac, indicating that these three different methods of analysis had been adequate in these cases.

From the data provided, the SCP concluded that genetically modified oilseed rape varieties derived from Liberator pHoe6/Ac are substantially equivalent to their conventional counterparts except for the introduced traits.