In vivo and ex vivo inflammatory responses of the esophageal mucosa to food challenge in adults with eosinophilic esophagitis

Elimination diets without the causative foods induce histological and clinical remission in patients with eosinophilic esophagitis (EoE), an allergen-driven type 2 inflammatory disease of the esophagus.1 However, current tests using skin or serum are poorly predictive of the causative foods,2 likely because the allergic inflammation may be restricted to the esophagus. We aimed to determine whether in vivo and ex vivo challenge of the esophageal mucosa with whole food extracts could yield clinically and immunologically relevant information about esophageal responses to specific foods. During endoscopy, the esophageal mucosa of 12 EoE patients was challenged by local injection of 3 common food triggers (cow's milk, wheat, and apple) and 3 foods based on patient's clinical history, and by local flush (i.e., spray) of a mixture of the 6 foods. Acute local responses were monitored for 20 min. Skin prick tests (SPT) and serum IgE measurements were also performed. Esophageal biopsies were exposed to foods in culture to analyze inflammatory mediator production, which was compared with 6 non-EoE controls. Methods are fully described in the Appendix S1. Patient characteristics are provided in Table 1. All patients were previously diagnosed with EoE and presented with the typical symptoms and endoscopic signs of EoE at time of endoscopy. Of the 11 patients that underwent injections, acute responses characterized by edema, erythema, or smooth muscle contraction (determined by the formation of a muscular ring) at the injection site were observed in 8 patients after injection with apple (n = 4), peanut (n = 4), wheat (n = 3), milk (n = 2), tomato (n = 1), egg (n = 1), and mango (n = 1) (Table 1; Figure 1A). In addition, after the end of the endoscopy, 4 patients experienced dysphagia, cramping retrosternal pain or burning sensation that was similar to pain occurring after ingesting those foods. Of the in total 17 foods that induced acute responses following local injections, 9 foods (53%) corresponded with patient's clinical history, 6 (35%) with SPT results and 6 (35%) with serum IgE results. The local flush with a mixture of foods also induced acute responses but, unlike the injections, these responses were barely notable and were observed in only 4 patients (Table S1). Our results confirm the observations of our previous study that esophageal food challenge can trigger local responses in adult EoE patients.3 However, there was no clear relation between foods that induced a response by mucosal injection, and SPT or serum IgE. The fact that the foods that induce mucosal responses do not necessarily show positive SPT and/or serum IgE results, and the fact that SPT and serum IgE are poorly predictive of the causative foods,2 indicate that local esophageal challenge may indeed be needed for a better prediction of the causative foods. Nonetheless, given the moderate responsiveness to challenge by flush, the clinical challenges associated with injections, and the invasiveness for patients as endoscopic challenge can induce short lasting but severe symptoms, both challenge tests will not likely become a useful test in clinical practice. In contrast, a less invasive biopsy-based ex vivo food challenge test may be considered a promising tool for the identification of causative foods in EoE patients. Non-challenged EoE esophageal biopsies maintained in culture for 24 h showed increased production of total IgE (13.7 vs. 0.1 ng/mg, p = .0002), IL-5 (12.5 vs. 1.1 pg/mg, p = .0288), IL-6 (29.8 vs. 1.5 ng/mg, p = .0047), IL-8 (86.6 vs. 23.2 ng/mg, p = .0069), IL-13 (28.6 vs. 0.0 pg/mg, p = .0080), and MCP-1 (659 vs. 112 pg/mg, p = .0320) compared with non-challenged biopsies from controls (Figure 1B). Eotaxin, IL-9, and IFN-γ were below the detection limit. Analysis of protein levels based on peak eosinophil count did not provide additional insights (data not shown). Furthermore, when exposing biopsies to food in culture, an immunological response is triggered that may reflect the inflammatory cascade seen in EoE. Interestingly, IL-5 levels were increased after ex vivo exposure to milk (89.8 vs. 12.5 pg/mg, p = .0195), and IL-9 was increased after exposure to apple (132.3 vs. 0.0 pg/mg, p = .0039; Figure 1C). To our knowledge, we are the first to report food-specific induction of IL-5, an important factor in eosinophil trafficking,4 and IL-9, a promotor of mast cell expansion and function,5 in the inflamed esophagus of EoE patients, highlighting a potential role for both cytokines in the allergen-specific immune response in EoE. Lastly, we used a machine learning approach6 to study whether the ex vivo challenge test can better discriminate clinically suspected (as provided in Table 1) from non-suspected foods than the conventional SPT and serum IgE. Indeed, the ex vivo challenge test outperformed SPT/serum IgE with an AUC of 0.64 vs. 0.5 (Figure S1), evidencing sufficient discriminative scores.7 Performing food re-challenges based on the ex vivo results was beyond the scope of this study. This study has limitations. Our study was conducted in a small cohort, and the tested foods were not proven by elimination diets. Extending the current study in a larger cohort of EoE patients in which causative and safe foods have been identified is needed to shed more light on the usefulness of the ex vivo test to identify causative foods and guide elimination diets. Furthermore, EoE is patchy in biopsies. Normalization of cytokine levels for epithelial/immune cell composition of the biopsies is therefore needed for standardization of the ex vivo test. In conclusion, we demonstrated that results of food challenge using esophageal tissue provide distinct results from tests using skin and serum and may better reflect clinical response to food exposure. Esophageal biopsy tissue culture is a functional model of EoE and could potentially be used as an ex vivo model for esophageal food challenge to (i) study the food-induced immune response and (ii) identify causative foods to guide elimination diets, and therefore warrants further validation and development. This research is funded within the Partnership between NWO domain Applied and Engineered Sciences and Danone Nutricia Research and with additional financial support from Topsector Agri and Food, project number 16495 with the acronym LOIRE. AJB is supported by Vidi grant 91718300 from the Netherlands Organization for Scientific Research NWO. AJB received research funding from Norgine, Thelial, SST and received speaker and/or consulting fees from Laborie, Arena, EsoCap, Medtronic, Dr. Falk Pharma, Calypso Biotech, Alimentiv, Sanofi, Reckett, Regeneron and AstraZeneca (all unrelated to this work). RvR received consultancy fees from HAL Allergy, Citeq, Angany, Reacta Healthcare, Mission MightyMe and AB Enzymes, speaker fees from HAL Allergy, ThermoFisher Scientific and ALK, and has stock option of Angany (all unrelated to this work). MTJVA is employed by Danone Nutricia Research. JG and BCAMVE are partly employed by Danone Nutricia Research. All other authors report no conflict of interest. Appendix S1: Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.