Standardization of double blind placebo controlled food challenge with soy within a multicentre trial

Background Multicentre trials investigating food allergies by double blind placebo controlled food challenges (DBPCFC) need standardized procedures, challenge meals and evaluation criteria. We aimed at developing a standardized approach for identifying patients with birch related soy allergy by means of DBPCFC to soy, including determination of threshold levels, in a multicentre setting. Methods Microbiologically stable soy challenge meals were composed of protein isolate with consistent Gly m 4 levels. Patients sensitized to main birch allergen Bet v 1 and concomitant sensitization to its soy homologue Gly m 4 underwent DBPCFC. Outcome was defined according to presence and/or absence of ten objective signs and intensity of eight subjective symptoms as measured by visual analogue scale (VAS). Results 138 adult subjects (63.8% female, mean age 38 years) underwent DBPCFC. Challenge meals and defined evaluation criteria showed good applicability in all centres involved. 45.7% presented with objective signs and 65.2% with subjective symptoms at soy challenge. Placebo challenge meals elicited non-cardiovascular objective signs in 11.6%. In 82 (59.4%) subjects DBPCFC was judged as positive. 70.7% of DPBCFC+ showed objective signs and 85.4% subjective symptoms at soy challenge. Subjective symptoms to soy challenge meal in DBPCFC+ subjects started at significantly lower dose levels than objective signs (p < 0.001). Median cumulative eliciting doses for first objective signs in DBPCFC+ subjects were 4.7 g [0.7–24.7] and 0.7 g [0.2–4.7] total soy protein for first subjective symptoms (p = 0.01). Conclusions We present the hitherto largest group of adults with Bet v 1 and Gly m 4 sensitization being investigated by DBPCFC. In this type of food allergy evaluation of DBPCFC outcome should not only include monitoring of objective signs but also scoring of subjective symptoms. Our data may contribute to standardize DBPCFC in pollen-related food allergy in multicentre settings. Trial registration EudraCT: 2009-011737-27.


Background
There is still debate whether allergen-specific immunotherapy (AIT) with birch pollen improves birch pollen-related food allergy [1,2]. One reason for partly contradictory data of previous trials on this topic is the lack of standardized tests to assess clinical reactions to birch pollen-related foods [1,2]. Double blind placebo controlled food challenge (DBPCFC) is considered as gold standard in diagnosis of food allergy [3]. When performing DBPCFC within multicentre trials, there is a need for standardization of challenge meals (CM), procedures and evaluation criteria [4][5][6][7][8]. In 2012, the PRACTALL group published a proposal for DBPCFC standardization [7], but interpretation of results still mainly depends on the investigator's judgment [9]. The definition of relevant and comprehensive cut off values for subjective [10,11] as well as for objective signs is challenging and differences between observers in interpreting DBPCFCs have been shown [12]. Another major problem is the interpretation of symptoms to placebo meals [10]. There exist only few published data on the frequency and type of those placebo reactions [13][14][15].
With regard to growing numbers of patients with birch pollen-related soy allergy, partly with severe reactions [16][17][18][19], we aimed at identifying subjects with birch pollen-related soy allergy by DBPCFC. Here we present a standardized approach for DBPCFC with newly developed challenge meals and evaluation scores that were set up for a multicentre trial. In contrast to challenge meals having been applied in other trials on birch pollenrelated food allergies, i.e. to apple or hazelnut [20,21], we expected best possible standardization of challenge meals due to the availability of purified soy proteins [22]. In a follow-up investigation patients with birch pollen-related soy allergy were to be included in a trial investigating any effect of birch pollen AIT on this type of food allergy [1].

Setting and patient selection
Between January 2010 and February 2013, birch allergic adults (18-65 years) were recruited in 16 centres (15 German, one Swiss). They underwent standardized allergy interview, skin prick test (SPT) with a panel of frequent respiratory allergens including birch (Allergopharma GmbH & Co KG, Reinbek, Germany) and serum IgE test for Bet v 1 and Gly m 4 (ThermoFisher, Freiburg, Germany). Eligibility for DBPCFC was defined at presence of ≥3.5 kU/l of specific IgE for birch allergen Bet v 1 and ≥0.7 kU/l for its soy homologue Gly m 4. Pre-challenge assessments ensured no interference of acute or chronic diseases or drugs. DBPCFC was preferentially performed in the morning to exclude as far as possible confounding factors as heavy meals (a light, allergen free breakfast was allowed) as well as physical or psychological stress. The trial (EudraCT: 2009-011737-27) was approved by the central ethical committee at Universität Leipzig, Germany (No. 230-09-ff-09112009), the local German boards and the local ethical committee Zurich (KEK-2010-0039). All patients gave written informed consent.

Challenge meal and DBPCFC
Soy and placebo CM consisted of soy free cocoa, carob flour, oat flakes, rice flour, and sugar. Nine doses of isolated soy powder Supro ® 760IP (88% protein, 12% ash, moisture, fat; Uelzena, Uelzen, Germany) were included in soy-containing CM. Single dose levels of soy protein were 0.0004-0.0044-0.05-0.15-0.5-1.5-2.5-5-15 g; resulting in a maximum cumulative dose 24.7 g protein (28.1 g powder). Placebo-CM also contained Sinlac ® (rice flour, carob flour, saccharose; Nestlé Nutrition GmbH, Frankfurt, Germany). Five volunteers without known type 1 sensitization to birch or soy participated in triangle test [4] to investigate any sensory differences between soy and placebo meals. For further validation of CM, patients were asked to guess wether they had received active or placebo meal at the last dose level applied. As patients with clinical reactions rather would suspect to have had active CM, we only considered those patients with negative outcome of DBPCFC. Trial sites prepared meals freshly by adding defined amounts of tap water to centrally produced powdered meals (Pharmacy of the University Hospital, Leipzig, Germany). At weeks 0, 6 and 12, all batches of soy meals (levels 4/9), and at week 0, all batches of placebo meals (level 9) and of soy protein isolate were analysed for Gly m 4 contents [22] as well as for microbiological contaminations (Institute of Laboratory medicine, Clinical chemistry and Molecular diagnostics, University Hospital, Leipzig, Germany). DBPCFC was performed on two separate days outside of birch pollen season (randomized sequence of soy and placebo meals). Meals were kept for several seconds in the oral cavity before swallowing. Increasing dose levels were applied at 20 min intervals.
Fully blinded investigators followed calculation/classification procedures and assessments of reactions independently from application of soy or placebo CM.
DBPCFC had to be stopped (i) at dose level 9, (ii) at occurrence of definite objective signs or (iii) on patient's demand. Lowest observed adverse effect level (LOAEL), and median cumulative threshold dose (MCTD) were determined in the study population with positive DBP-CFC. A positive reaction to soy or placebo meal was defined if a patient presented at least one of the following three reaction types: Objective type-an objective sign as defined (01-10) and/or Single subjective type-at least a single subjective symptom as defined (S1-S8) with VAS value reaching 1.5 cm or more and/or Subjective sum type-two or more subjective symptoms (S1-S8) with single VAS values each reaching 0.5 cm or more and summing up to 4 cm or more in total.
Any objective sign or subjective symptom occurring at placebo meal was classified as either minor or major. Only major placebo reactions (MPR) were taken into account when defining outcome of DBPCFC as positive (+) or negative (−) (including patients with major placebo reactions, Fig. 1). Criteria of MPR were fulfilled, when the intensity of objective signs or subjective symptoms to placebo exceeded those to soy meal (Table 1).

Serology
All subjects were investigated for total IgE and sIgE against Bet v 1 and Gly m 4 (ThermoFisher, Freiburg, Germany) before DBPCFC. 56/82 DBPCFC+ patients being eligible for AIT were further investigated for sIgE against Bet v 2, Gly m 5 and Gly m 6 (ThermoFisher).

Statistics
Between-group comparisons of DBPCFC+ and DBP-CFC− patients were performed descriptively by exact Fisher's test and Mann-Whitney U test depending on the scale and distribution (binary or ordinal/skewed) of DBPCFC characteristics after soy challenge. In DBP-CFC+ patients, symptom characteristics after soy and  Table 1 Page 4 of 10 Treudler et al. Clin Transl Allergy (2016) 6:39 placebo challenge were compared by McNemar test and Wilcoxon paired rang sum tests. P-values < 0.05 were considered as statistically significant. No adjustment for multiple tests was performed since the aspects investigated primarily served as additional data description.
Coefficients of variation at or below 20% were found for the predefined shelf life of 12 weeks in all series measured. No Gly m 4 was-detectable in any placebo batch with respect to the detection limit of the ELISA used [22]. No significant microbiological impurities were detected in any batch within these 12 weeks. Preparation and application of CM was easy to handle in all centres. Subjective Any single SubjS* (S1-8) at lower dose level compared to soy 5 or below/any VAS placebo * 6/any VAS soy * Single* Any single SubjS* (S1-8) at same dose level like any single SubjS* to soy 5/any VAS placebo * 5/any VAS soy * Single** Any single SubjS** (S1-8) at same dose level like any single SubjS to soy and no difference of at least 2 cm in favor of soy 4/any VAS placebo ** (e.g. 0.6 cm) 4/any VAS soy *** < VAS*** placebo -2 cm (e.g. 2 cm)

Study
Sum type*** Two or more aggregated SubjS** (S1-10) and two or more aggregated SubjS** (S1-8) to soy (any but same dose level for both) and no difference between sum VAS values of at least 2 cm in favor of soy  (Figs. 2 and 3).
In 82/138 (59%) subjects, DBPCFC was judged as positive. Detailed data on objective signs and subjective symptoms in DBPCFC+ and DBPCFC− subjects as well as numbers of major placebo reactions are given in Table 2. Subjective placebo reactions (single and/or sum values) were reported by 41% (n = 57). Characteristics of objective signs and subjective symptoms of DBPCFC+ and DBPCFC− subjects are given in Fig. 2 and 3. Significant differences between DBPCFC+ and DBPCFC− patients were demonstrated with regard to occurrence of objective signs and subjective symptoms ( Table 2).  (Fig. 2). No objective signs occurred in 24 (29.2%) subjects at soy CM and in 71 (86.6%) at placebo CM.

Characteristics of patients with positive DBPCFC
Ten (12.2%) of DBPCFC+ subjects patients reported MPR of subjective type to placebo (

Table 2 Characteristics of DBPCFC positive and DBPCFC negative subjects
Maximum dose levels applied and occurrence of objective signs, subjective symptoms and major placebo reactions in DBPCFC positive and DBPCFC negative patients at soy and placebo challenge meals. 95% confidence intervals are given (per cent based on the number of patients per group; in case of non-overlapping confidence intervals significant differences with α = 5% between populations exist) n.s. not significant a For definition of Major placebo reactions see

Discussion
Here we present the hitherto largest group of adults with combined sensitization to birch allergen Bet v 1 and to its soy homologue Gly m 4 being investigated by DBPCFC. In this multicentre trial, 82 subjects (59.4%) had positive DBPCFC according to harmonized evaluation criteria. With the aim of identifying DBPCFC positive patients being eligible for a follow-up trial on AIT against birch pollen allergen, we decided to only include patients with defined IgE cut off values, though being aware that IgE values cannot safely predict reactivity at DBPCFC [25]. As clinical symptoms may occur in Bet v 1 sensitized subjects already at first consumption of soy products [26], our inclusion criteria did not ask for history of clinical allergy. Since no recipe was available at the planning stage of our trial, newly developed challenge meals were composed as microbiologically stable desserts based on powdered ingredients. Soy protein levels of active meals were chosen on the basis of previously published data [27].
Details on DBPCFC evaluation criteria used in clinical trials are rarely reported in the current literature and no consensus exists on how to discriminate positive from negative outcomes. Presentation of detailed evaluation criteria used in this trial in comparison with criteria of PRACTALL [8] and EuroPrevall [27] is given in Table 3.
In our DBPCFC+ patients, objective signs were most frequently of the mucocutaneous type. At placebo challenge, fifty percent (15/30) of all objective signs were due to heart rate increase and drop of blood pressure. Both may be symptoms of immediate type reaction but may as well be induced by psychovegetative factors. Thus, as a single parameter, neither of these cardiovascular symptoms seemed appropriate for identifying DBPCFC+ patients. Also, flush, rhinitis and conjunctivitis, possibly against the background of underlying atopic diseases, were more frequently seen in patients judged as DBP-CFC− than in DBPCFC+ patients at placebo challenge.
Regarding subjective symptoms at DBPCFC, there is no consensus on how to monitor or how to score. In our trial as well as in a recent report [29], visual analogue scales (VAS), were used. We predefined cut off values for data analyses with regard to experiences from pain measurement [23,24]. As VAS values are prone to a certain risk of being unspecific, we requested a VAS value of 1.5 cm or more to be classified as a single positive reaction. Most of subjective symptoms documented in our trial were typical for contact urticaria of the oropharyngeal sites [17,18].
For evaluation of pollen-related food allergy with frequent predominance of subjective symptoms [17,18], our evaluation criteria may be more comprehensive than those proposed by PRACTALL [8] which neither consider in detail mucocutaneous symptoms nor include measuring of symptom's degree. However, we are aware that the complex definition of major subjective placebo reactions presented for this trial may not be suitable for routine use.
It was recently shown, that standardized food challenges are subject to a great variability in interpretation subjective symptoms (SubjS, either single or sum type) in 70/82 (85.4%) DBPCFC+ patients, respectively. SubjS to soy challenge meal started at significantly lower dose levels than ObjS (p < 0.001). Cumulative threshold dose eliciting objS and subjS in 50% of subjects (ED 50 ) can be extrapolated of clinical symptoms [9]. Regarding placebo events, in the current literature, no systematic documentation in DBPCFC in adults, like in our trial, has been published. In children, objective signs and/or subjective symptoms occurred in at least 2.8% [12] or 12.9% [14] of challenges. We documented an overall number of objective placebo reactions in 21.6% which highlightens the complexity of DBPCFC evaluation.
DBPCFC is usually considered positive when objective signs occur exclusively on active and not on placebo challenge [8,28,29]. This definition is insufficient in birch pollen related food allergy were patients often suffer from subjective symptoms only and/or may present objective signs against their atopic background not being induced by DBPCFC. We suggest that minor placebo reactions should be compatible with positive DBPCFC outcome supposed that there are relevant differences between reactions at active and at placebo meal. With the aim of harmonization in a multicentre setting, we therefore suggested a definition of relevant (major) placebo reactions (Table 1).
According to theses definitions, we identified MCTD for subjective symptoms (0.7 g) and for objective signs (4.7 g), which are, however, not considered representative for all subjects with birch related soy allergy due to selected inclusion criteria. Nevertheless, data were close to those determined in a previous trial on soy allergy with MCTD of 0.9 (subjective) and 4.8 g (objective), respectively [27].
Due to our experience we feel that evaluation of DBP-CFC outcome, only can be standardized to a certain extend. Final decision upon relevance of occurring sign and symptoms during DBPCFC still relies on the clinical investigator, especially with regard to circumstances during the challenge procedure as well as considering any physical and/or psychological comorbidities of the patient.

Conclusions
We present a standardized approach for DBPCFC with soy that includes application of challenge meals with stable Gly m 4 values as well as determination of threshold levels in a multicentre setting. For evaluation of any treatment effects on birch pollen-related food allergy, we see an urgent need not only for providing standardized challenge meals but also for investigating validity and reliability of DBPCFC outcome scoring systems taking into account intensity of objective clinical signs as well as subjective symptoms.  . TH reports personal fees (payment for organising lectures) from Akademie Fresenius and non-financial support (travel/accommodation/meeting expenses unrelated to activities listed) from European Academy of Allergology and Clinical Immunology, from Deutscher Allergie-und Asthmabund e.V. and from International Life Science Institute/Health and Environmental Sciences Institute. SV reports personal fees from Food Allergy Resource and Research Program, Lincoln, NE USA, personal fees from Medical University of Vienna, Austria, grants from Monsanto Company, personal fees from American Academy of Asthma, Allergy and Immunology, personal fees from Deutsche Dermatologische Gesellschaft, personal fees from Westdeutsche Arbeitsgemeinschaft für pädiatrische Pneumologie und Allergologie e.V., Koln, Germany, personal fees from Gesellschaft für pädiatrische Allergologie und Umweltmedizin, personal fees from Ärzteverband Deutscher Allergologen, personal fees from Swiss Society for Allergy and Immunology, personal fees from Schattauer Allergologie Handbuch, personal fees from Elsevier Nahrungsmittelallergien und Intoleranzen, personal fees from Karger Food Allergy: Molecular Basis and Clinical Practice, non-financial support from German Research Foundation, non-financial support from Federal Institute for Risk Assessment, nonfinancial support from European Directorate for the Quality of Medicines and Health Care, nonfinancial support from European Academy of Allergy and Clinical Immunology, non-financial support from Deutscher Allergie-und Asthmabund, nonfinancial support from Association Monegasque pour le Perfectionnement des Connaissances des Médicins, non-financial support from Federal Office of Consumer Protection and Food Safety, non-financial support from German Chemical Society (GDCh), non-financial support from AKM Allergiekongress, non-financial support from International Union of Immunological Societies, outside the submitted work. Other authors declare no conflicts of interest.

Ethical standards
The trial (EudraCT: 2009-011737-27) was approved by the central ethical committee at Universität Leipzig, Germany (No. 230-09-ff-09112009), the local German boards and the local ethical committee Zurich (KEK-2010-0039). All patients gave written informed consent to participate.