For many years the practical difficulties of obtaining meaningful information from conventional toxicology studies on the safety of whole foods have been well recognized (OECD, 1996). The limitations of conventional toxicological studies became particularly apparent when animal feeding studies were used to assess the safety of irradiated foods.
Animal studies are a major element in the safety assessment of many compounds such as pesticides, pharmaceuticals, industrial chemicals and food additives. In most cases however, the test substance is well characterized, of known purity, of no particular nutritional value and human exposure is generally low. It is therefore relatively straightforward to feed such compounds to animals at a range of doses, some several orders of magnitude greater than the expected human exposure levels, in order to identify any potential adverse health effects of importance to humans. In this way it is possible, in most cases, to determine levels of exposure at which adverse effects are not observed, and so set safe upper limits by the application of appropriate safety factors.
By contrast, foods are complex mixtures of compounds characterised by wide variation in composition and nutritional value. Due to their bulk and effect on satiety they can usually only be fed to animals at low multiples of the amounts that might be present in the human diet. In addition, a key factor to consider in conducting animal studies on foods is the nutritional value and balance of the diets used, to try to avoid the induction of adverse effects which are not related directly to the material itself. Detecting any potential adverse effects and relating these conclusively to an individual characteristic of the food can therefore be extremely difficult. Another consideration in deciding the need for animal studies is whether it is appropriate to subject experimental animals to such a study if it is unlikely to give rise to meaningful information.
In practice, very few foods consumed today have been subject to any toxicological studies, yet they are generally accepted as being safe to eat. In developing a methodology for the safety assessment of new foods, it was essential to establish a benchmark definition of safe food. This was taken up by OECD in 1991 who said that food is considered safe if there is reasonable certainty that no harm will result from its consumption under anticipated conditions of use.
The difficulties of applying traditional toxicological testing and risk assessment procedures to whole foods, meant that an alternative approach was required for the safety assessment of genetically modified foods. This led to development of the concept of substantial equivalence.
This approach acknowledges that the goal of the assessment is not establishing absolute safety but to consider whether the genetically modified food is as safe as its traditional counterpart, where such a counterpart exists.
The Consultation agreed that the practical difficulties already identified in relation to the application of conventional toxicology studies to whole food preclude their use as a routine safety assessment technique for genetically modified foods. The Consultation also recognised that the use of toxicology studies could not be justified from an animal welfare perspective where it was unlikely to result in meaningful information. In addition, the Consultation 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 concept of substantial equivalence was developed as a practical approach to the safety assessment of genetically modified foods. The Consultation agreed that substantial equivalence should be seen as a key step in the safety assessment process. The application of the concept is not a safety assessment in itself; it does not characterize the hazard, rather it is used to structure the safety assessment of a genetically modified food relative to its conventional counterpart. As a starting point, the genetically modified organism (plant, micro-organism or animal), and/or foods derived from it, is compared with its closest traditional counterpart in order to identify any intended and unintended differences which then become the focus of the safety assessment. Data for comparison should be obtained using validated methods and analyzed using appropriate statistical techniques. The comparative approach should take into account agronomic, genetic and chemical aspects and only when all of these have been considered can an objective assessment of safety be made. The type and extent of further studies depend on the nature of the differences and whether or not they are well characterised. Studies should be carried out in accordance with good laboratory practise.
4.2.1 Basic Principles
Several international organisations have already addressed the issues associated with the safety assessment of novel foods and, in the present context, genetically modified plants and micro-organisms (WHO, 1991; OECD, 1993; WHO, 1995; FAO, 1996; ILSI, 1996; Commission of the European Communities, 1997). It is generally agreed that such an assessment requires an integrated and stepwise, case-by-case approach and some authorities have developed decision trees to assist in determining the extent of testing required in specific cases (SCF, 1997; UK ACNFP, 1995). This approach is useful in determining appropriate safety assessment strategies.
The safety assessment of a genetically modified food is directed by the results of a comparison between the genetically modified food and its conventional counterpart. It follows a stepwise process aided by a series of structured questions. Factors taken into account in the safety assessment include:
effects of processing/cooking;
the recombinant DNA (e.g. stability of insertion, potential for gene transfer);
protein expression product of the novel DNA;
effects on function;
possible secondary effects from gene expression or the disruption of the host DNA or metabolic pathways, including composition of critical macro-, micro-nutrients, anti-nutrients, endogenous toxicants, allergens, and physiologically active substances; and,
potential intake and dietary impact of the introduction of the genetically modified food.
The above factors are particularly pertinent to the assessment of foods derived from genetically modified plants. When assessing the safety of foods derived from genetically modified animals and micro-organisms, other factors may need to be taken into account on a case-by-case basis.
4.2.2. Need for Animal Studies
If the characterization of the food indicates that the available data are insufficient for a thorough safety assessment, animal testing may be deemed necessary. This would particularly be the case if the food were expected to make a significant dietary contribution, if there is no history of consumption of the novel gene product or if the modification affects several metabolic pathways.
In the situation where the genetically modified food differs from the traditional counterpart by the presence of one or a few new genes and their products, it may be possible to isolate and study these in a manner analogous to conventional toxicity testing of food additives. However it is essential to ensure that the material tested is biochemically and functionally equivalent to that produced in the genetically modified food. This provides the possibility of increasing the sensitivity of toxicity tests compared with that possible if the products of the genetically modified plants had been fed directly and avoids some of the artefacts that can occur in toxicity tests on whole foods. However, this strategy is only applicable if the preceding detailed analysis does not reveal significant changes other than those expected. Otherwise testing of the whole food may be required. When animal testing is conducted on the whole food, it should generally be on the food as consumed by humans. The type of animal study would need to be considered on a case by case basis. In addition to investigating potential toxicological effects, animal studies may also be required if the genetic modification directly or indirectly affects the content or bioavailability of nutrients.
Where toxicological studies are considered necessary to assess the safety of long term consumption of a food in the diet, it is generally considered that a sub-chronic study of 90-days duration is the minimum requirement to demonstrate the safety of repeated consumption of a food in the diet. This may need to be preceded by a pilot study of short duration to ensure that the diet is palatable to the test species and that the levels of incorporation of the test article are appropriate, e.g. the control diet containing the equivalent level of the comparator does not produce effects, as a result of normal levels of natural toxicants present in traditional foods accepted as safe. The highest dose level used in any animal study should be the maximum achievable without causing nutritional imbalance while the lowest level used should be comparable to the anticipated human intake.
The need for additional toxicological tests should be considered on a case-by-case basis taking into account the results of the 90-day study and other studies. For example, proliferative changes in tissues during the 90-day study may indicate the need for a longer-term toxicity study.
Conventional toxicological tests are of limited value in assessing whole foods, including genetically modified foods. Based on the maximum levels of the whole food that can be incorporated into experimental diets as indicated previously, a margin of safety may be estimated based on the absence or nature of adverse effects and likely human exposure. Improved experimental designs should take into account the need for nutritionally adequate animal diets, avoiding some of the inappropriate testing of foods or products.
It has been suggested that the use of biomarkers of early effects might increase diagnostic value and sensitivity of toxicity tests on foods (Schilter et al., 1996). However, it will be necessary not to confuse adaptive and toxic effects in applying this approach.
In achieving the objective of conferring a specific target trait (intended effect) to the host organism by the insertion of defined DNA sequences, additional traits could, theoretically, be acquired or existing traits lost (unintended effects). The assessment of genetically modified foods involves methods to detect such unintended effects and procedures to evaluate their biological relevance and their impact on food safety.
Unintended effects may be due to factors such as random insertion events which might result in disruption of existing genes, modifications of protein expression or formation of new metabolites. The expression of enzymes at high levels may give rise to secondary biochemical effects, e.g. an altered metabolic flux resulting in changed metabolite patterns.
The potential occurrence of unintended effects is not specific to the use of recombinant DNA techniques. Rather, it is an inherent and general phenomenon that can occur in conventional breeding. One of the approaches adopted to cope with this problem is to select/ discard plants with unusual and undesired phenotypic and agronomic parameters at an early stage of the plant variety development. The practice of consecutive back-crossing is also a common procedure used to eliminate unintended effects.
Unintended effects due to genetic modification may be subdivided into two groups: those which are "predictable" based on metabolic connections to the intended effect or knowledge of the site of insertion and those which are unexpected. Due to the increased precision of genetic modification compared to conventional breeding, it may become easier to predict pathways likely to be influenced by unintended effects.
The comparator used to detect unintended effects should ideally be the near isogenic parental line 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 chosen. The resulting natural variation should be taken into account in assessing the statistical significance of the unintended effect.
Where statistically significant unintended differences are observed, their biological significance should be assessed. This may be assisted by knowledge of the mechanisms leading to the changes. In order to assess the biological and safety relevance of an unintended effect, 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.
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 considered as useful alternatives (non-targeted approach). 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.
In the future, genetic modifications of plants are likely to be more complex perhaps involving multiple between-species transfers and this may lead to an increased chance of unintended effects. Where differences are observed using profiling techniques, the possible implications of the differences with respect to health need to be considered.
The Consultation acknowledged that the concept of substantial equivalence had attracted criticism. This criticism relates, in part, to the mistaken perception that the determination of substantial equivalence was the end point of a safety assessment rather than the starting point. Further disagreement may have arisen from reference to three outcomes of substantial equivalence discussed previously (i.e. substantially equivalent, substantially equivalent apart from defined differences, and not substantially equivalent) (FAO, 1996).
Having considered the way in which the concept of substantial equivalence is currently used, and the possible use of alternative strategies, the Consultation concluded that application of the substantial equivalence concept contributes to a robust safety assessment framework. The Consultation was satisfied with the approach used to assess the safety of the genetically modified foods that have been approved for commercial use.
It was agreed that communication of the principles involved in safety assessment could be improved. The Consultation concluded that the key message to be conveyed is that substantial equivalence is a concept used to identify similarities and differences between the genetically modified food and a comparator with a history of safe food use which subsequently guides the safety assessment process.
The Consultation reiterated that a consideration of compositional changes was not the sole basis for determining safety. Safety can only be determined when the results of all aspects under comparison are integrated.
It was recognised that whole foods do not lend themselves to the standard safety evaluation principles (WHO 1987) used for food additives and other chemicals and that a quantitative assessment of risk of individual whole foods from whatever source cannot be achieved. The Consultation agreed that assessing safety relative to existing foods offered the best means of assessing the safety of genetically modified foods.
The Consultation considered the issue of long term effects from the consumption of genetically modified foods and noted that very little is known about the potential long term effects of any foods. In many cases, this is further confounded by wide genetic variability in the population, such that some individuals may have a greater predisposition to food-related effects.
In this context, the Consultation acknowledged that for genetically modified foods, the pre-marketing safety assessment already gives assurance that the food is as safe as its conventional counterpart. Accordingly it was considered that the possibility of long term effects being specifically attributable to genetically modified foods would be highly unlikely. Furthermore, it was recognised that observational epidemiological studies would be unlikely to identify any such effects against a background of undesirable effects of conventional foods. Experimental studies, such as randomised controlled trials (RCTs), if properly designed and conducted, could be used to investigate the medium/long term effects of any foods, including genetically modified foods. Such studies could provide additional evidence for human safety, but would be difficult to conduct. In this respect, it is also important to recognise the wide variation in diets and dietary components from day to day and year to year.
The Consultation was of the view that there were presently no alternative strategies that would provide a better assurance of safety for genetically modified foods than the appropriate use of the concept of substantial equivalence. Nevertheless, it was agreed that some aspects of the steps in the safety assessment process could be refined to keep abreast of developments in genetic modification technology. New methodologies, such as profiling techniques, offer the means of providing a more detailed analytical comparison. However, it was recognised that much more developmental work was necessary before such methods could be validated.