In 1996, the International Food Biotechnology Council and the Allergy and Immunology Institute of the International Life Sciences Institute (IFBC/ILSI) presented a decision-tree approach to the evaluation of the potential allergenicity of the novel gene products (proteins) in genetically modified foods (Metcalfe et al., 1996). This allergy assessment strategy has been widely adopted by the agricultural biotechnology industry. It is a strategy which focuses on the source of the gene, the sequence homology of the newly introduced protein to known allergens, the immunochemical binding of the newly introduced protein with IgE from the blood serum of individuals with known allergies to the transferred genetic material, and the physicochemical properties of the newly introduced protein (Metcalfe et al., 1996; Taylor, 1997).
In the 1996 Joint FAO/WHO Consultation on Biotechnology and Food Safety, the issue of allergenicity of genetically modified foods was specifically addressed for the first time. An assessment approach similar to that developed by IFBC/ILSI was advocated that included the following criteria: source of the transferred genetic material, molecular weight, sequence homology, heat and processing stability, effect of pH and/or gastric juices (digestive stability), and prevalence in foods. The 1996 Consultation concluded that a rational scientific approach to the assessment of the allergenicity of genetically modified organisms can and should be undertaken as part of the overall safety assessment approach. Furthermore, the 1996 Consultation made several recommendations relative to allergenicity of genetically modified foods:
The transfer of genes from commonly allergenic foods should be discouraged unless it is documented that the gene transferred does not code for an allergen.
Foods found to contain an allergen transferred from the organism which provided the DNA should not be considered for marketing approval unless such products can be clearly identified in the marketplace and this identity will not be lost through distribution and processing. Further, that labelling approaches may not be practical in these situations, and that particular problems exist for consumers who cannot read, or who may not be provided with labels.
Involved organizations should consider the appropriateness of, and/or actions to take, in respect to foods containing new protein(s) that are determined to have the characteristics of an allergen, even though no patient population is known to exist which has an allergy to this gene product.
The identification of food allergens and the characteristics of these allergens that define their immunogenicity be encouraged.
In the 2000 Joint FAO/WHO Consultation on Safety Aspects of Genetically Modified Foods of Plant Origin, the issue of the allergenicity of genetically modified foods was specifically addressed again. The IFBC/ILSI decision-tree approach was adapted, with minor changes, for the evaluation of novel proteins introduced into genetically modified foods (Annex 3). The said Consultation concluded that if a genetically modified food contains the product of a gene from a source with known allergenic effects, the gene product should be assumed to be allergenic unless proven otherwise. The transfer of genes from commonly allergenic foods should be discouraged unless it can be documented that the gene transferred does not code for an allergen. The novel proteins introduced into genetically modified food should be evaluated for allergenicity on the basis of the decision-tree shown in Annex 3. The 2000 Consultation noted that the IFBC/ILSI decision tree as adapted by FAO/WHO in Annex 3 had received some criticism related to certain of the criteria involved in the decision tree. The 2000 Consultation further concluded, additional criteria should be considered for the addition to the decision-tree approach when the source of the genetic material is not known to be allergenic. The level and site of expression of the novel protein and the functional properties of the novel protein would be two such criteria.
The 2000 Joint FAO/WHO Consultation recommended that WHO/FAO should be encouraged to convene an Expert Consultation on the assessment of the allergenicity of genetically modified foods and the novel proteins contained therein. The Consultation should focus on the development of an improved decision-tree approach for the assessment of the allergenicity of genetically modified foods and on the standardization/validation of specific criteria, such as optimal methods for assessment of digestive stability. With this background, the current consultation undertook efforts to develop an improved decision-tree approach using as a start, the existing IFBC/ILSI decision tree as adapted by the 2000 FAO/WHO Consultation (Annex 3) .
5.2. The FAO/WHO 2001 Decision Tree
After consideration of the current status of scientific information and extensive discussion, the Consultation developed a new decision tree (Annex 4) that will be referred to throughout the remainder of this report as the FAO/WHO 2001 decision tree. This new decision tree builds upon previous approaches to examining allergenicity but also encompasses several additional strategies.
5.3. Food containing a gene derived from a source known to be allergenic
When the expressed protein comes from a source known to be allergenic, the analysis presented in the FAO/WHO 2001 decision tree focuses on both sequence homology and subsequent assessment of potential allergenicity of the expressed protein with sera of patients allergic to the source material (Annex 4). Sequence homology is the initial step to be performed. Criteria for a positive outcome in the analysis of sequence homology are reviewed in Section 6.1. When sequence homology to a known allergen is demonstrated, the product is considered allergenic, and no further testing is typically undertaken. If no sequence homology to a known allergen is demonstrated, specific serum screening for the expressed protein is undertaken. These investigations focus on assessment of the possible allergenicity of the expressed protein using sera from patients allergic to the source material (Section 6.2). These patients should be carefully defined according to international guidelines. If the patients donating sera have a low level of sensitisation, the usefulness of those sera in exhibiting reactivity to the expressed protein may be compromised. Therefore, it is suggested to include only patients with a level of sensitisation to the allergen source of more than 10 kIU/L of specific IgE.
In contrast to previous decision-tree strategies, the FAO/WHO 2001 decision tree makes no distinction between commonly and less commonly allergenic source materials with respect to specific serum screening. Thus, specific serum screening is undertaken irrespective of the relative frequency of allergy to the source material in question, provided sera are available (Section 6.2). Insufficient information exists in the literature supporting an increased risk of a severe reaction for patients with hypersensitivity to commonly allergenic foods as opposed to less commonly allergenic foods.
The degree of confidence in the results of the specific serum screening will depend upon the number of sera that are available for analysis. To achieve 95 % certainty that a major allergen (a major allergen is defined as one to which more than 50 % of individuals sensitive to that substance react in IgE-specific immunoassays) from the source material has not been transferred, a negative result must be obtained with at least 6 relevant sera. To achieve 99 % certainty that a major allergen from the source material has not been transferred, a negative result must be obtained with at least 8 relevant sera. To achieve 99.9 % certainty that a major allergen from the source material has not been transferred, a negative result must be obtained with at least 14 relevant sera. Furthermore, by using 17 relevant test sera, a 95 % probability exists of detecting a minor allergen (a minor allergen is defined as one to which less than 50 % of individuals sensitive to that substance react in IgE-specific immunoassays) from the source to which at least 20 % of the affected population are reactive. By using 24 relevant sera, a 99 % probability exists of detecting a minor allergen from the source to which at least 20 % of the affected population are reactive. An argument can be made for using fewer sera if relevant sera are not available, but this modified approach carries the risk of a false negative outcome. However, the use of larger numbers of sera is advocated, whenever possible, to increase the confidence associated with negative immunoassay results as described above. The Consultation also recognizes that the use of a smaller number of very well documented, high quality sera may be preferable to the use of larger numbers of lesser-quality sera. The in vitro method applied should be a validated assay measuring specific IgE (Section 6.2).
Any positive outcome defines the product as likely allergenic, and will normally lead to discontinuation of product development. A negative outcome of the specific serum screening prompts further analysis using targeted serum screening (Section 6.3), pepsin resistance (Section 6.4) and animal models (Section 6.5) (see Annex 4). Additionally, in vivo/ex vivo testing in allergic patients may also be appropriate in circumstances where confirmation of positive results in the specific serum screening is wished; or where a negative outcome of appropriate in vivo/ex vivo testing would be more convincing than a positive outcome of the specific serum screening provided that well documented allergic subjects were used in the in vivo/ex vivo testing. The ex vivo/in vivo methods include skin prick testing (Bruijnzeel-Koomen et al, 1995), basophil histamine release (Bindslev-Jensen and Poulsen, 1996) and oral challenge (Bock et al, 1988; Bruijnzeel-Koomen et al, 1995). It is anticipated that these procedures will require approval from Ethics Committees (Internal Review Boards). Therefore, the FAO/WHO 2001 decision tree does not include human in vivo testing as a mandatory tool, but in vivo testing may be considered in selected cases.
An equivocal outcome of the specific serum screening would lead to further analysis using targeted serum screening, pepsin resistance or animal models (see Annex 4). Again, ex vivo/in vivo testing involving patients allergic to the source material may also be considered.
The FAO/WHO 2001 decision tree is not applicable to the evaluation of foods where gene products are down-regulated for hypoallergenic purposes. In such cases, in vivo testing including skin prick testing, open challenges, and double-blind, placebo-controlled food challenges would be required.
5.4. Food containing a gene derived from a source not known to be allergenic
When the expressed protein comes from a source that is not known to be allergenic, the FAO/WHO 2001 decision tree focuses on (1) sequence homology to known allergens (food and environmental), (2) targeted serum screening for cross-reactivity with sera from patients allergic to materials that are broadly related to the source material for the gene, (3) pepsin resistance and (4) immunogenicity testing in animal models (Annex 4). In this situation the search for homologous allergens is based on two procedures.
The first step is a database search for an allergen with a homologous amino acid sequence, according to the principles described in Section 6.1. If this search reveals a level of homology with a known allergen that suggests a potential for cross-reactivity, the expressed protein is considered to be an allergenic risk. No further evaluation for allergenicity would typically be necessary.
The second step is conducted if no such homologous protein is found. In such cases, cross-reactivity is tested with a panel of serum samples that contain high levels of IgE antibodies with a specificity that is broadly related to the gene source (Section 6.3). For this targeted serum screen, 6 groups of source organisms are distinguished: yeast/moulds, monocots, dicots, invertebrates, vertebrates and others. A panel of 50 serum samples with high levels of IgE to allergens in the relevant group is used to search for IgE antibodies that are cross-reactive with the expressed protein. If a positive reaction is obtained with one of these sera, the expressed protein is considered to be an allergenic risk and further evaluation for allergenicity would typically not be necessary. If a gene were obtained from a bacterial source, no targeted serum screening would be possible, since no normal population of individuals are known to be sensitised (IgE mediated) to bacterial proteins.
When a positive outcome is obtained in targeted serum screening, further evaluation using in vivo/ex vivo approaches as described in Section 5.3 may be conducted if desired to seek confirmation of the results of the targeted serum screening. If the results obtained with in vivo/ex vivo testing differ from those obtained with targeted serum screening, these results would be more convincing than a positive outcome in the targeted serum screening provided that appropriate, well documented allergic subjects were used in the in vivo/ex vivo testing.
If no cross-reactive serum is found, the protein is analysed for pepsin resistance and for evidence of immunogenicity in appropriate animal models according to the protocols provided in Sections 6.4 and 6.5.
5.5. Post marketing surveillance
The Consultation acknowledges that the pre-market allergenicity assessment of the genetically modified food gives a satisfactory safety assurance. However, it is recognised that due to the wide genetic variability in the human population and different geographical dietary intake, further evaluation for adverse effects of the genetically modified food should be considered once the product has reached the market. This could provide additional safety assurance.
Ideally, a notifying, self-reporting system for any adverse health effects, both for consumers and for employees in the food production industry should be put in place. Reported data should be validated with respect to:
the clinical outcome in relation to allergenicity
the causality between the reported adverse effect and the specific genetically modified food/food ingredient exposure
These validated data should be recorded, consolidated and published. Such a system could benefit from experiences of existing national surveillance systems (e.g. disease control centres, poisoning centres).
However, the feasibility of post-marketing surveillance systems should be further explored, since there are number of problems to be addressed, including:
traceability and labelling of the genetically modified food/food ingredient
lack of background data on prevalence and incidence in food related allergies
existence of many confounding food and non-food related factors
changes in diets over time
lack of trained experts and infrastructure, especially in developing country settings
5.6. Other Criteria that were Considered
5.6.1. Level of expression
Highly allergenic proteins are often expressed at relatively high levels. However, allergens can sensitize susceptible individuals at less than milligram levels, possibly at less than microgram levels (Sorva et al., 1994; Jarvinen et al., 1999). The elicitation of objective symptoms in already sensitized individuals can also occur at low levels of exposure, but has not been documented below 500 micrograms (Rance and Dutau, 1997; Hourihane et al., 1997). It is therefore not possible to define a level of expression below which a protein can be considered safe from the allergenicity point of view. Thus, level of expression cannot yet be incorporated into the assessment of the allergenicity of genetically modified foods.
5.6.2. Unintended effects
In achieving the objective of conferring a specific target trait (intended effect) to the host organism by the insertion of DNA sequences, additional traits could, theoretically, be acquired or existing traits lost or enhanced (unintended effects). Unintended effects may be due to factors such as random insertion events, which might result in disruption of existing genes and modification of protein expression. While unintended effects are not specific to the use of recombinant DNA techniques, any such effects should be identified to the maximum extent possible and their impact upon the allergenicity of the genetically modified food should be assessed.
With respect to allergenicity, two types of unintended effects could be envisioned. First, the gene insert may activate or suppress existing host genes in an inordinate fashion leading to either over-expression or under-expression of specific proteins. If the host plant contains known allergenic proteins, then the possibility that the levels of these allergens has been elevated should be considered as part of the safety evaluation process. Secondly, if evidence is obtained from comparison of the genetically modified food to its conventional counterpart that the insertion of the gene creates additional new proteins, then these proteins should be evaluated for their potential allergenicity using the approach described herein.
 in vivo (using
allergic human subjects)/ex vivo (using cells or tissue culture from allergic