The basophil activation test differentiates between patients with wheat‐dependent exercise‐induced anaphylaxis and control subjects using gluten and isolated gluten protein types

Abstract Background Oral food challenge using gluten and cofactors is the gold standard to diagnose wheat‐dependent exercise‐induced anaphylaxis (WDEIA), but this procedure puts patients at risk of an anaphylactic reaction. Specific IgE to ω5‐gliadins as major allergens and skin prick tests to wheat may yield negative results. Thus, we designed a proof‐of‐principle study to investigate the utility of the basophil activation test (BAT) for WDEIA diagnosis. Methods Different gluten protein types (GPT; α‐, γ‐, ω1,2‐ and ω5‐gliadins, high‐molecular‐weight glutenin subunits [HMW‐GS] and low‐molecular‐weight glutenin subunits [LMW‐GS]) and gluten were used in different concentrations to measure basophil activation in 12 challenge‐confirmed WDEIA patients and 10 control subjects. The results were compared to routine allergy diagnostics. Parameters analyzed include the percentage of CD63+ basophils, the ratio of %CD63+ basophils induced by GPT/gluten to %CD63+ basophils induced by anti‐FcεRI antibody, area under the dose‐response curve and test sensitivity and specificity. Results GPT and gluten induced strong basophil activation for %CD63+ basophils and for %CD63+/anti‐FcɛRI ratio in a dose‐dependent manner in patients, but not in controls (p < 0.001, respectively). BAT performance differed from acceptable (0.73 for LMW‐GS) to excellent (0.91 for ω5‐gliadins) depending on the specific GPT as evaluated by the area under the receiver operating characteristic curve. Patients showed individual sensitization profiles. After determination of the best cut‐off points, ω5‐gliadins and HMW‐GS showed the best discrimination between patients and controls with a sensitivity/specificity of 100/70 and 75/100, respectively. Conclusion This study shows the alternative role of BAT in better defining WDEIA and the causative wheat allergens. The best BAT parameters to distinguish WDEIA patients from controls were %CD63+ basophil values for ω5‐gliadins and HMW‐GS.


| INTRODUCTION
Wheat-dependent exercise-induced anaphylaxis (WDEIA) is a rare, but potentially life-threatening cofactor-induced wheat allergy. ω5gliadins and high-molecular-weight glutenin subunits (HMW-GS) are most often reported as major allergens, but reactions to other gluten protein types (GPTs) from wheat, like low-molecular-weight glutenin subunits (LMW-GS), α-and γ-gliadins were also described. All GPT together constitute gluten, the storage proteins in wheat flour. [1][2][3][4] WDEIA diagnosis is challenging, because of the variety of possible allergenic wheat proteins and the combination with a cofactor. Skin prick tests (SPTs) and specific IgE (sIgE) to wheat may be negative. Even sIgE to ω5-gliadins are only positive in about 80% of WDEIA patients, indicating that other GPT also play a role in WDEIA. Wheat product and exercise challenge failed to induce symptoms in the majority of patients despite a clear history. 1,4,5 Thus, oral food challenge with gluten, that has a protein content of 70%-80% compared to the 8%-15% in wheat flour, combined with cofactors is often needed to overcome non-responsiveness. 6,7 A new approach to complement WDEIA diagnosis is the basophil activation test (BAT) combined with florescence-activated cell sorting. Stimulation with an allergen-containing solution (allergen test solution [ATS]) induces upregulation of the expression of cell surface proteins, such as CD63 or CD203c. BAT-derived parameters such as the percentage of basophils that respond to a given dose of the ATS or the area under the curve (AUC) of a dose-response curve have been shown to be sensitive biomarkers corresponding to the clinical severity of anaphylactic reactions. 8,9 The BAT has been established for the identification of different immediate allergies, like allergy against wheat, 8,10 hymenoptera venom, 11 and alpha-galactose. 12 The aim of this proof-of-principle study was to investigate the utility of the BAT to improve WDEIA diagnosis and to determine individual sensitization profiles in WDEIA patients to different isolated and well-characterized GPT and gluten.

| Study population
The following exclusion criteria were considered in the selection of participants to avoid potential confounding factors and/or health risk to any of the participants: Pregnancy/lactation; systemic intake of corticosteroids (cortisone) 3 weeks and/or antihistamines (anti-pruritic drugs) 1 week before the start of the test; intake of laxatives, anti-diarrhea drugs, thyroid hormone preparations, antibiotics, immunosuppressive drugs, analgesic drugs (aspirin, NSAIDs, etc.) taking psychotropic drugs and certain blood pressure medications (ACE inhibitors, ß-blockers); serious internal diseases (gastrointestinal, neurological, cardiovascular, rheumatic diseases, celiac disease, cancer, kidney diseases, acute infections, etc.); bronchial asthma.
A total of 23 participants were consecutively recruited from the medical center (15 f, 8 m, 25-76 years). Twelve of them were patients with a history of WDEIA based on positive oral food challenge, SPT, sIgE, and clinical history (5 f, 7 m, 26-60 years, Table 1).
Provocations had been done with 8-32 g gluten intake as described. 7,13 Some patients, depending on their history, were given increasing doses of cofactors (500-1000 mg of ASA ± 10-20 ml of 95% ethanol; Braun, Melsungen, Germany diluted with 200 ml of black currant-flavored water) 30 min before gluten challenge and standardized aerobic and anaerobic exercise was undertaken 30-60 min after gluten ingestion. Eleven individuals without a history of any wheat-related disorder were included in the study as controls (10 f, 1 m, 25-76 years, Table 2).
The study protocol was approved by the ethics committee of the Technical University of Munich and all participants gave written informed consent before being included in the study.

| Skin prick test
SPT was carried out on the forearm with the following substances: wheat flour, gluten, isolated LMW-GS, HMW-GS, and gliadins.
A 10% histamine-dihydrochloride solution (ALK-Abello, Hørsholm, Denmark) was used as positive and isotonic sodium chloride solution (Fresenius Kabi Deutschland GmbH) as negative control. 7 Details of the production and characterization of GPT can be found in the Supporting Methods.

| Preparation of BAT ATSs
Gliadins were extracted from wheat gluten using 60% aqueous ethanol. After dialysis and lyophilization, the gliadin fraction was separated into ω5-, ω1,2-, α-, and γ-gliadins by preparative reversedphase high-performance liquid-chromatography. The glutenins were extracted from the residue after gliadin removal using 50% aqueous propanol, 60°C and reducing conditions. The HMW-GS and LMW-GS were obtained by sequential precipitation with 40% and 80% acetone, respectively. 14-16 Details of the production and characterization of the GPT can be found in the Supporting Methods. ATS were prepared and stored at −20°C in aliquots until use in BAT.

| Basophil activation test
For quantitative determination of in vitro basophil activation, Flow CAST (Buehlmann Laboratories AG) was used, as described previously. 12 Venous blood was collected from participants in EDTA tubes and used immediately. The blood samples were gently homogenized at room temperature (RT). Per measurement, 50 μl of ATS (concentra-

| Study population
Twelve patients (7 m, 5 f; age range: 26-60 years; median age: 48 years) with a clinical history of WDEIA and positive challenge test were included in the study ( Table 1). The control population consisted of 11 controls without a clinical history of any wheat-related disorder; six subjects were atopic. One subject with atopy was excluded, because of non-responsiveness to the positive control anti-FcɛRI mAb in the BAT. Therefore, 10 controls were analyzed further (1 m, 9 f; age range: 25-76 years; median age: 44 years) ( Table 2). There were no significant differences in wheal or erythema diameter between wheat flour, gluten, gliadins, HMW-GS, and LMW-GS between the patients (p > 0.05).

| SPT, serum sIgE and total IgE
Significantly higher values were found for total IgE and sIgE

Note:
The results are the median of a double determination n = 2 (except WDEIA patient 4, n = 1). Isotonic sodium chloride was used as negative (NaCl) and a 10% histamine solution as positive control (histamine). The diameters for wheals and erythema were documented in mm. A result is classified as positive (marked in bold), when the diameter of the wheal, caused by a test substance, is greater or equal than the diameter of the wheal caused by the negative control with 3 mm added. The range and median of all patients' results per test substance is documented.
GABLER ET AL.

| Evaluation of the response induced by ATSs in BAT
The induced allergenic response to the ATS was evaluated by three  Table 4 shows the AUCs of dose-response curves of gluten and GPT for patients and controls.

| ROC curves
The ROC curve describes how accurately the test can distinguish patients from controls. The greatest AUC values for concentrationindependent ROC curves were determined for %CD63 + basophils as characteristic, for ω5-gliadins (0.908), HMW-GS (0.867), and gluten (0.850) ( Table 5; Figure 3). Concentration-independent ROC curves were generated from the maximum values for %CD63 + basophils out of all tested concentrations for each single ATS in patients and controls. The optimal discrimination threshold (cut-off) for %CD63 + basophils, when a basophil activation is classified as "allergen response" to a ATS, was determined for best sensitivity and specificity of concentration-independent ROC curves ( Table 5).
Concentration-dependent ROC curves showed best results at 4.00 mg/ml for ω5-gliadins and HMW-GS, 2.00 mg/ml for LMW-GS and gluten as well as 0.8 mg/ml for ω1,2-, α-and γ-gliadins. More information about ROC curves and results for concentrationdependent ROC curves are presented in the Supporting Information (Tables S1 and S2; Figure S2).
In this proof-of-principle study we show that BAT with gluten and GPT  Abbreviations: c, control; HMW-GS, high-molecular-weight glutenin subunits; LMW-GS, low-molecular-weight glutenin subunits; p, patient.
T A B L E 5 Patient and control data from concentration-independent ROC curves for ATSs from gluten and ω5-, ω1,2-, α-, and γ-gliadins and HMW-GS/ LMW-GS with AUC and optimal discrimination threshold for %CD63 + basophils (cut-off), when a basophil activation is classified as "allergen response" to an ATS CD63 and CD203c are both in use as activation markers in BAT, while CD63 is the most common one. The upregulation of CD63 is closely associated with basophil degranulation induced by allergen stimuli. 19 Hoffmann et al. (2016) reported that the upregulation of CD203c also occurs to non-degranulation stimuli, which is not the case for CD63. 20  In the present study, a well-characterized representative gluten sample was used to isolate single GPT, α-, γ-, ω1,2-, and ω5-gliadins as well as HMW-and LMW-GS. Detailed information about the basophil activation in patients to GPT were obtained. 23 The challenge of poor solubility of gluten proteins in aqueous solutions was overcome by increasing their solubility and accessibility via partial hydrolysis with pepsin. 24 The highly specific anti-FcεRI mAb that imitates bridging of the receptor by an allergen has been used as a positive control in BAT for numerous years. Rubio  Sensitivity and specificity of basophil activation to these substances at optimal cut-off in WDEIA patients as compared to atopic and nonatopic control subjects were good for ω5-gliadins (sensitivity:
The maximum %CD63 + basophils turned out to be the best parameter to differentiate between patients and controls, with significant differences for all tested allergens. It is conspicuous that sensitivity and specificity were higher for ω5-gliadins, HMW-GS and gluten than for ω1,2-, α-, and γ-gliadins, and LMW-GS, because ω5-gliadins and HMW-GS have previously been identified as most relevant allergens in patients with WDEIA. [1][2][3][4] Matsuo et al. recommended to determine sIgE against epitopes of ω5-gliadins and HMW-GS in combination for WDEIA diagnosis. 27 Based on our results, we can also confirm this recommendation for their use in BAT. Other allergenic GPT were less important in our study. 2,4 BAT identified the sensitization profile of WDEIA patients to be particularly directed against ω5-gliadins and HMW-GS, but in individual patients also against α-gliadins, γ-gliadins, LMW-GS, and ω1,2-gliadins. For example, two patients (p6, p9) showed high responses to LMW-GS 59.8 (p9) and 31.8 (p6) %CD63 + basophils, concentration 4.0 mg/ml). [1][2][3][4] One limitation of our study is the comparatively small number of WDEIA patients and controls. Due to the very low prevalence of F I G U R E 3 Concentration-independent receiver operating characteristic curves for ω5-gliadins, high-molecular-weight glutenin subunits (HMW-GS), and gluten, which had the highest sensitivity and specificity. The maximum basophil activation %CD63 + basophils out of all tested concentrations for each single allergen test solution in patients and controls was taken to generate ROC curves GABLER ET AL. According to the results of the proof-of-principle study we showed the potential of the BAT as alternative to routine SPT and sIgE measurements in WDEIA diagnosis. The BAT turned out to be promising to study the allergenicity of different GPTs, which becomes only possible after special preparation to increase water solubility, as required for BAT. Our findings indicate the use of %CD63 + basophils as best parameter to discriminate between patients and controls and highlight the allergenicity particularly of ω5-gliadins and HMW-GS for WDEIA.