Novel strategies for predicting allergenicity: development of a ranking method and screening tools to assess the allergy risk of innovative proteins

Abstract

Abstract To protect individuals who already have or are at risk of developing immune‐mediated adverse reactions to food, novel foods and genetically modified organisms (GMOs) undergo an allergenicity risk assessment. There are shortcomings in this process that could be improved through use of well‐defined clinically relevant allergen molecules with different allergenic potential. The objective of this project was to develop novel strategies for predicting allergenicity of innovative/novel proteins that address this issue. We undertook a systematic review of allergen molecules in foods listed on Annex II of the Food Information for Consumers Regulation together with additional foods known to cause IgE‐mediated food allergies in at least one European region with a prevalence of 0.5%. Around 750 in‐scope papers were quality assessed to allow clinical relevance of allergen molecules to be ranked. The best characterised clinically relevant allergens were identified in peanut, hazelnut, cow's milk, fish and crustacean shellfish with data lacking for allergens from foods such as pecan, Macadamia, lupin and melon. Furthermore, an assessment of in silico tools allergenicity prediction found that, whilst many were able to correctly predict allergenicity, none were able to provide an output that could be linked to the clinical relevance. Building on these outcomes an approach for allergenicity risk assessment has been developed that brings together elements of exposure assessment, combining in silico, in vitro, and in vivo methods. Tools for assessment of risks of cross‐reactive allergies are more mature and only require refinement to improve the outputs to inform the allergenicity risk assessment process. However, as mechanisms underlying development of food allergy are not fully elucidated, and remain a matter of ongoing research, prediction of de novo sensitisation is uncertain.