A Brown Norway rat sensitisation model can rank foods according to their allergenicity in humans

Increasing introduction of novel foods, especially from alternative protein sources, increases the need for reliable tools to assess allergenicity, particularly for assessing de novo sensitising capacity. Currently, there are no tools to predict de novo sensitisation, and as in silico and in vitro approaches cannot capture the complexity of the sensitisation process nor the associate clinical relevance, in vivo animal models remain the best option for developing such predictive tool. Thus, in this study, an adjuvant-free intraperitoneal Brown Norway rat model was developed to evaluate allergenicity and rank the sensitising capacity of six different well-known foods according to their allergenicity in humans and subsequently transferred to a CBA/J mouse model. Using a reference set of foods, ranging from low- to medium- to high-allergenicity, the rat model demonstrated clear and reproducible differentiation between foods, particularly when optimal dose and immunisation regimes were selected. Integration of multiple endpoints, including humoral (IgG1, IgE), functional (basophil activation), and clinical responses (ear swelling test and oral food challenge), enabled the assessment of sensitisation and supported the development of a composite sensitisation capacity ranking system with machine learning. Overall, the Brown Norway rat model was shown to be sensitive, with reproducible responses capable of distinguishing between foods with varying allergenicity. This study highlights the importance of selecting an appropriate experimental design and suitable endpoint parameters for assessing sensitisation and provides a promising step towards developing predictive tools for evaluating sensitising capacity of novel foods, although further validation using a larger set of reference foods is required.