Safety Assessment Testing Procedures for GM Foods

The substantial equivalence concept uses a comparative approach by comparing food derived from genetically engineered plants with their conventional counterparts with a history of safe use and is considered the first step in the safety assessment process. The concept takes into account both intended and unintended changes that occur in the GE plant or food derived from it. This comparison provides the basis on which to focus any further toxicological requirements for a safety assessment. Due to the difficulties of applying traditional toxicological testing and risk assessment procedures, used for testing single chemicals, a different approach is required for the safety assessment of foods derived from plants including recombinant DNA plants. However, feeding studies that use whole foods rather than isolated compounds may be appropriate in experimental animals when there are significant compositional changes in the food derived from recombinant-DNA plants (CAC, 2003).

To evaluate the safety of food derived from genetically engineered plants, the first step should be to determine an appropriate comparator with a history of safe use. The comparator is usually a traditional crop/food or other edible variety from the same species from which that GM crop/food was developed and preferably the direct parental line (Kuiper et al., 2004). Two scenarios may be considered (European Commission, 1997): Firstly, the novel food is equivalent to an accepted traditional food or ingredient, in which case no further testing is needed. Secondly, the novel food is equivalent to the traditional counterpart except for some well defined differences, or food differs from the traditional counterpart in multiple and complex respects; such a novel food would require assessment targeted to these differences.

If there are no significant differences between the genetically engineered crop and the comparator or if there are differences that will, within reason, not adversely affect health, the genetically engineered product is considered as safe as its counterpart. Currently hazard identification tests routinely employed in the safety assessment for foods derived from genetically engineered plants typically include compositional analysis, toxicological and allergenicity testing (FAO/WHO, 2001). Nutritional assessment, though not a safety issue in itself is also carried out where appropriate to detect any significant unintended nutritional changes or intended nutritional modification that could affect the nutritional status of individuals consuming the food (FAO, 2008). Notably, nutritional assessment can be quantified through compositional analysis.

As part of the safety testing for crops containing transgenes, studies are undertaken to investigate the biochemical composition of tissues that may be components of food or feed. Compositional analyses are uniquely required by regulatory agencies for all transgenic crops, even though genetic changes to transgenic crops are generally less complicated and better characterized than more traditional crop improvement methods (Cockburn, 2002; SOT, 2002; Crawford, 2003).

Food composition is concerned with both beneficial and harmful components in the human diet including nutrients, bioactive non-nutrients, antinutrients, toxicants, contaminants and other potentially useful and dangerous elements. Based on the comparative approach of the concept of substantial equivalence, compositional data are used as a starting point to guide the subsequent direction of the safety assessment process based on whether there are significant deviations in composition that would warrant further safety investigation (FAO/WHO, 2000).

Due to the numerous biochemical constituents in living organisms, a subset of components has been selected for inclusion in composition studies (Oberdoerfer et al., 2005; McCann et al., 2005; Obert et al., 2004). The Organization of Economic Cooperation and Development (OECD) and the International Life Science Institute (ILSI) have developed scientific consensus documents on the safety of novel foods and feeds detailing key food and feed nutrients, antinutrients, toxicants and allergens for a variety of crops to be considered in compositional analysis (OECD, 2001-2008; ILSI, 2003).

Key nutrients or key antinutrients in a particular food are defined as those that may have a substantial impact in the overall diet. They are generally major constituents namely proteins, carbohydrates, fats as nutrients or enzyme inhibitors as antinutrients or minor components like minerals and vitamins. Key toxicants are those toxicologically significant compounds known to be inherently present in the plant, such as glycoalkaloids in potatoes, glycosides in cassava, and high levels of selenium in wheat. Another key component that is also targeted for testing is allergens (FAO/WHO, 2008).

In evaluating the safety of transgenic crops through compositional comparison, the concentration of each component in the transgenic crop is compared to the concentrations in non-transgenic counterparts which are preferably the parental lines. In the absence of the parental lines, the OECD advocates the use of several different control lines to determine whether any observed differences fall within the range of conventionally derived varieties (OECD, 1993).

It is critical that the transgenic crop and its non-transgenic control are grown side by side to capture the potential effects of growing conditions (Herman et al., 2007). It is also essential to take into account the compositional differences that are found among the same variety grown under different environmental conditions (Reynold, 2005) and during different climatic conditions (e.g. different growing seasons). Statistical analyses are usually performed on data for the genetically modified variety and the conventional counterpart, leading to acceptance or rejection of the null hypothesis that they are substantially similar with a certain probability (Kuiper et al., 2001). Due to limited information on the composition of some minor plant species, especially with regard to the antinutrient and natural toxin profiles, compositional analysis data should be used in conjunction with data from other tests to screen for unintended effects of genetic modification. For further reading on composition analysis, see the References section.

The approach used for the assessment of potential toxicity in foods involves biochemical characterization of the new product from the inserted DNA element. The new substances can be conventional components of plant foods such as proteins, fats, carbohydrates, or vitamins. Potential toxicity of substances that have not previously been consumed in food is assessed on case-by-case basis. The type of studies to be performed on these substances may include studies on metabolism, toxicokinetics, sub-chronic toxicity, chronic toxicity, reproduction and development toxicity according to the traditional toxicological approach using animal models. The methodologies for oral toxicity studies have been developed in international forums (OECD, 1995; US FDA, 2003; EFSA, 2008).

In the case of proteins that are the product of genetic modification of a plant through genetic engineering, the assessment of potential toxicity is carried as follows;

(i) Determination of amino acid sequence similarity between known protein toxins and antinutrients and that of the introduced protein. For these comparisons, public domain database are used including GenBank, EMBL, PIR and Swiss Prot. Similarity to a known toxin could lead to toxicological testing to address the potential impact of the homology. However, it should be noted that no standard methodology has been adopted for establishing how significant the “similarity to a known toxin” should be before a concern is raised.

(ii) In vitro digestibility studies are performed to determine the resistance of the novel product to acid by simulating the conditions in gastric and intestinal (GI) fluids. Proteins that are broken down in the GI system are more likely to be safe following oral consumption than those that resist digestion if for no other reason than they are unlikely to retain biological activity following degradation (Delaney et al., 2008).

(iii) Stability of the protein to heat and/or processing is also assessed.

(iv) Acute (14 days) oral toxicity studies based on animal models. In the case of single proteins, acute tests should be sufficient since laboratory animals have been shown to exhibit acute toxic effects from exposure to proteins known to be toxic to humans (Sjoblad et al., 1992). If acute toxicity tests are not conclusive, there may be a need for additional subchronic (between 3 and 12 months) and chronic (long-term) toxicity testing (FAO/WHO, 2000).

If a novel protein is found to have no significant sequence similarities to known protein toxins, is readily digested in conditions that mimic mammalian digestion, is not stable to heat and/or processing and either has a history of safe human consumption and/or does not cause any toxic effect in acute toxicity testing then it can be reasonably concluded that the protein is non-toxic to humans. However, if a novel protein fails one or more of the criteria discussed above then further investigation may be required and conclusions made using the weight of evidence approach (FAO/WHO, 2000). In carrying out these tests good laboratory practices (OECD, 1998) should be applied and the studies conducted according to internationally recognized guidelines (OECD, 1995; NRCILAR, 1996).

As part of the assessment of the potential toxicity of a novel protein, it is important to also determine if the activity of the novel protein in the organism is likely to produce secondary effects. If other substances are found to accumulate as a result of the activity of a novel protein for example, it is also important to include an assessment of the potential toxicity of such substances using the traditional isolated or single compound testing methods (CAC, 2003).

In addition to the tests discussed, subchronic whole food studies may also be carried out by feeding GE food, typically to rats or mice for at least 28 days (Kuiper et al., 2001). Subchronic and chronic testing in animals are not currently routinely carried out as part of the safety assessment process (IM and NRC, 2004). For further reading on toxicological studies, see the References section.

A food allergy is a reaction of the immune system to an otherwise harmless food or food component. A major allergen is defined as that which elicits a specific Immunoglobulin E (IgE) antibody response in more than 50% of affected individuals (Taylor and Hefle, 2002). Overall, approximately 90% of all food allergies are associated with a small number of specific proteins represented by eight major allergenic foods: peanuts, tree nuts, cow’s milk, hen’s eggs, fish, crustacean, wheat, and soybeans (Metcalfe et al., 1996).

In the safety assessment of food produced through agricultural biotechnology, the assessment of the potential allergenicity of the novel proteins introduced into these foods is one of the key issues. Since almost all allergens are proteins (Bush and Hefle, 1996), there exists a possibility that any novel food protein might be an allergen. On the other hand, since so few of the proteins in nature are allergenic, the probability that a novel protein will become a food allergen is rather low. Nevertheless, the possibility of the presence of an allergen has to be investigated on a case by case basis. Unfortunately there is no single predictive test for the potential allergenicity of any novel protein (Taylor and Hefle, 2002). For this reason a weight of evidence approach is typically employed that comprises multiple tests to improve the predictive capability of the overall assessment.

The strategy focuses on the following key elements; (i) source of the gene, (ii) A comparison of the sequence homology of the newly introduced protein to known allergens, as is done with known toxins and antinutrients, (iii) the immunoreactivity of the novel protein with serum IgE from individuals with known allergies to the source of the transferred DNA, (iv) the resistance of the novel protein to digestion with pepsin and, (v) the immunogenicity of the novel protein in appropriate animal models (may be done but not a general requirement because no such test are validated).

If the gene source is known to be allergenic, it must be assumed that this gene encodes for an allergen from that source unless data are generated to disprove that assumption. More than 160 other foods and food-related substances have been associated with allergenic reactions in individuals (Hefle et al., 1996). All these sources and any other sources associated with environmental or occupational allergies would be classified as known allergenic sources under the FAO/WHO.

According to the FAO/WHO, if the gene source is known to be allergenic, the potential allergenicity of novel gene product can be determined with a reasonable degree of certainty using the specific screening test of sera from sensitive individuals. If on the other hand, the gene is obtained from a source with no history of allergenicty, then a specific serum screening test is obviously not possible.

A comparison of the amino acid sequence of the novel protein to the amino acids sequences of known allergens is a useful initial step in the determination of its allergenic potential regardless of the source of the gene (FAO/WHO, 2001). Typically, sequence homology searches comparing the structure of novel proteins to known allergens in a database (e.g. PIR, SwissProt and TrEMBL) are conducted using various algorithms such as FASTA (Pearson and Lipman, 1988; Ladics et al., 2007) to predict overall structural similarities.

As recommended by FAO/WHO (2001), a significant sequence homology occurs when there is a match of at least eight contiguous (a linear sequence of amino acids) identical amino acids or an overall homology of 35% or greater over a segment of 80 or more amino acids (CAC, 2003). If the sequence homology test is positive, the conclusion can be reached that the novel protein might be allergenic. However, confirmation of the results of the sequence homology tests can be sought with specific serum screening. This may be advisable when one of the sequence homology tests appears to yield a false positive result, such as an irrelevant match of six contiguous amino acids (Taylor and Hefle, 2002).

Sequence homology searches have limitations. In particular, comparisons are limited to the sequence of known allergens in publicly available databases and the scientific literature. There are also limitations in the ability of such comparisons to detect non-contiguous (the folded three dimensional form of the protein as opposed to linear section) epitopes capable of binding specifically with IgE antibodies (FAO/WHO, 2001). Epitope here refers to the part of a protein that is recognized by the immune system specifically by antibodies, B cells or T cells.

When the novel gene is obtained from a known allergenic source or the search for sequence homology identifies a match with a known allergenic source, a specific serum screening (in vitro immunoassay) is recommended where sera are available (FAO/WHO, 2001). A negative result in in immunoassays may not be considered sufficient, but should prompt additional testing, such as the possible use of ex vivo protocols. An ex vivo vitro procedure is described as testing for allergenicity performed using cells or tissue culture from allergenic human subjects (FAO/WHO, 2001). However, from a practical point of view, product developers avoid using genes from known allergenic sources that would trigger allergenicity concerns and require extensive tests to discount any possibility of allergenicity.

In specific serum screening, the novel protein is bound to a solid phase and then blood serum from individuals known to be allergenic is used to determine whether the allergen-specific IgE antibodies in the serum will react with epitopes on the novel protein (Taylor and Hefle, 2002). The sera to be used for testing should be well characterized and standardized criteria should be used for selection of allergenic individuals (Eigenmann and Sampson, 1997). It is also imperative that during assay development, conditions be optimized in order to minimize false-positive IgE binding (Ladics, 2008).

Proteolytic stability is seen as a useful criterion in the assessment of the protein’s potential to become a food allergen. This is because it has been observed that there exists a correlation between resistance to digestion by pepsin and allergenic potential (Astwood et al., 1996). In simulated gastric and intestinal digestive models, known food allergens exhibited greater proteolytic stability than known non-allergenic food proteins. Many of the novel proteins introduced into food through agricultural biotechnology are also rapidly digested in these same model systems (Astwood et al., 1996).

The FAO/WHO weight of evidence approach advocates use of resistance to proteolysis with pepsin as a comparative measure of digestive stability for novel proteins introduced in foods through agricultural biotechnology. Digestive proteolysis in humans is individually variable and therefore the digestive test may not be fully predictive of the proteolytic stability of food allergens. However, the comparative resistance to proteolysis with pepsin is considered a reasonable comparative measure in the allergenicity measurement. Care should be taken so that a standardized protocol (e.g. FAO/WHO developed protocol) is used so that comparative data can be obtained.

Animal models are based on the assessment of induced antibody (i.e., IgE or IgG) responses and the frequency of responders in the test groups. Several models have been proposed using rats (Atkinson et al., 1996; Knippels et al., 1998; Knippels et al., 2000), mice (Dearman et al., 2000; Dearman and Kimber, 2001), swine (Helm et al., 2002), and dogs (Buchanan et al., 1997). Presently, none of the animal models have been tested with a wide range of allergens and non-allergens, and there is a lack of data on the reproducibility and predictive value (sensitivity and specificity) of any of the models (Ladics, 2008). A number of relevant questions still remain and include: (1) what is the most appropriate endpoint or design for an animal model? (2) What constitutes the measurement of a positive allergic response? (3) What is the most appropriate route of exposure or form of protein to test (i.e., isolated pure protein vs. protein in the food matrix). Research is currently on going to extensively evaluate and validate animal models (Ladics, 2008) for use in predicting the allergenicity of foods. Until such validation is possible the use of animal models cannot play a central role in the assessment of human allergenicity.