Peanut traces in packaged food products consumed by allergic individuals: Results of the MIRABEL project

Journal of Food Composition and Analysis 44 (2015) 196–204

Contents lists available at ScienceDirect

Journal of Food Composition and Analysis journal homepage: www.elsevier.com/locate/jfca

Original Research Article

Peanut traces in packaged food products consumed by allergic individuals: Results of the MIRABEL project Jutta Zagon a,*, Joerg Dittmer a, Chabi Fabrice Elegbede b, Alexandra Papadopoulos b, Albert Braeuning a, Ame´lie Cre´pet b, Alfonso Lampen a a

Federal Institute for Risk Assessment (BfR), Department Food Safety, Max-Dohrn-Str. 8-10, D-10589 Berlin, Germany French Agency for Food, Environmental and Occupational Health & Safety (ANSES), Risk Assessment Department (DER), 27-31 avenue du Ge´ne´ral Leclerc, 94701 Maisons-Alfort Cedex, France

b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 8 June 2015 Received in revised form 7 August 2015 Accepted 10 August 2015 Available online 18 August 2015

In the frame of the French research project MIRABEL, 899 food samples which contained no peanut ingredients according to the labeling list were analyzed for the presence of peanut allergen traces. Samples covered a broad range of products from ten major food categories. In a stepwise procedure, samples were screened using a sensitive lateral flow assay with a limit of detection of 2 ppm total peanut or 0.5 ppm peanut protein. Positive as well as suspect samples (139/899) were confirmed by real-time PCR with the same sensitivity. Positives in both approaches were quantified by two different commercial ELISA tests. 1% (9/899) out of all samples did contain measurable peanut DNA and protein traces above the detection limit of the applied methods. Six samples had a content of total peanut protein <5 mg/kg, two samples contained between 8 and 10 mg/kg and one sample a maximum of about 20 mg/kg. An excellent correlation was found between Ct-values obtained by PCR and ppm peanut calculated by ELISA. It is concluded that, in the light of future thresholds for labeling of relevant allergens, the methods used for peanut detection in this study are able to detect contaminations as low as 2 ppm. ß 2015 Elsevier Inc. All rights reserved.

Keywords: Peanut (Arachis hypogaea) detection Food analysis Food allergen Food composition Real-time PCR ELISA Lateral flow assay MIRABEL project

1. Introduction Peanut (Arachis hypogaea) is one of the fourteen major food allergen species that have to be labeled when used as ingredients in the recipes of packaged and non-prepackaged foods, according to the European legislation (European Directive 1169/2011/EC). Peanut is, like soybean, a member of the leguminous plant family and may elicit severe reactions in sensitized persons. It has long been known that peanut consumption is a frequent cause of lethal anaphylactic reactions (Yunginger et al., 1988; Sampson et al., 1992). The reported prevalence of allergies against peanut depends on the country and on the age of patients under study, but it is roughly

Abbreviations: CTAB, cetyltrimethyl ammonium bromide; COV, coefficient of variation; DNA, desoxyribonucleic acid; ED, effective dose; ELISA, enzyme-linked immunosorbent assay; LFA, lateral flow assay; LOAEL, lowest observed adverse effect level; LOD, level of detection; PCR, polymerase chain reaction; SD, standard deviation. * Corresponding author. Tel.: +49 (0)30 18412 3876; fax: +49 (0)30 18412 63876. E-mail address: [email protected] (J. Zagon). http://dx.doi.org/10.1016/j.jfca.2015.08.006 0889-1575/ß 2015 Elsevier Inc. All rights reserved.

estimated to be about 0.5–1.1% in the US population and also in some European countries (EFSA, 2004). Among food allergens, peanut is associated with the highest prevalence, which is estimated to be 0.3–0.75% of the French population (Morisset et al., 2005). It appears that children are more frequently and increasingly affected than adults (Grundy et al., 2002; Hourihane, 2011). Combining the observations of Rance´ et al. (2005) and Moneret-Vautrin (2008), peanut allergy prevalence in France is estimated to be 0.3% in adults aged 18 to 79 years, but 0.6% of children aged 3 to 17 years. Children are particularly endangered since peanut may be hidden in chocolate, snacks and biscuits. Once manifested, peanut allergy tends to persist for the whole lifetime (Husain and Schwartz, 2012). Peanut allergic adults and in particular parents of sensitized children are confronted with the problem of carefully avoiding any unintended contact with a potentially life-threatening allergen. A continuously updated epidemiological literature survey spanning from 1994 until present is published on the ‘allergome’ database website (www. allergome.org). In the frame of the French research project MIRABEL (Cre´pet et al., 2015) led by the French Agency for Food, Environmental and

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

Occupational Health & Safety (ANSES), about nine hundred food samples not labeled for peanut as an ingredient were collected from the French market and investigated for the presence of peanut traces. Aims of the project were (i) to evaluate the consumption behavior and attitude respecting product price and labeling of peanut allergic patients, (ii) to analyze the real market situation regarding the presence and amount of unintended peanut traces for labeled (with precautionary labels related to allergen traces on package such as ‘‘may contain traces of peanut’’) and unlabeled products, (iii) to assess and quantify the all over risk against the background of analytical results, medical data and individual consumption data, based on a Bayesian statistical model, and (iv) finally to derive a cost-/benefit analysis and management strategies from all data (Cre´pet et al., 2015). The investigation of 899 market samples for the presence of peanut traces, which is presented in this work, was an essential core part of the integrated framework MIRABEL, addressing aims (ii) and (iii) of the MIRABEL project outlined above. To meet the real life situation, products were not taken randomly but ranked into different categories and subcategories considering their labeling types (labeled/unlabeled) to reflect consumption habits of peanut allergic patients, as derived from the MIRABEL consumption survey. In order to optimize the sample plan to monitor allergen traces in products consumed by allergic patients, a Bayesian network was developed and applied in MIRABEL project (Elegbede et al., 2015). Samples were allocated to ten groups of food: breakfast cereals; cereal bars; bread and bakery products; appetizer products; pizzas; cream desserts, mousse or fresh desserts; biscuits and pastry; chocolate bars or chocolate spread; other chocolate products; and ice cream and sorbets. These ten major food categories were further divided into subcategories according to their ingredients and flavouring (supplementary information, Table A). Moreover, the brands of the collected products were the major ones consumed by the allergic patients in the MIRABEL consumption survey. A cascade of methods was applied for efficient analysis of the high number of samples. In the first step, all samples were screened for the presence of peanut with a sensitive and rapid immunological lateral flow assay (LFA). Secondly, positive as well as ambiguous and suspect samples were confirmed using a realtime PCR method with the same sensitivity as the LFA. Finally peanut traces were quantified with two different commercial ELISA (enzyme-linked immunosorbent assay) kits. 2. Material and methods 2.1. Food samples Between January and March 2013, 899 food samples were collected on the French market according to a previously developed and recently published model (Elegbede et al., 2015). The full sample list including the number of investigated lots is presented in the supplementary information (supplementary information, Table A). Due to the sample collection some slight differences appear between the optimized number of samples provided by the Elegbede et al. (2015) model and the number of samples actually collected: the recommended number of 900 samples was reduced by one item (one appetizer product ‘Curly Fromage’), because peanut was indicated as a regular ingredient on the product label. Additionally, some products were not well classified into the right categories by the patients and were thereafter reallocated to the appropriate category. That is why, for the categories ‘cream desserts, mousse or fresh desserts’ and ‘ice creams and sorbets’, larger differences can be observed between the optimized number of samples as provided by the model and the number of samples actually collected. According to

197

Elegbede et al. (2015) the optimal sample size for the category ‘cream desserts, etc.’ would have been 53 (versus 30 in this study) and for the category ‘ice cream and sorbets’ 37 (versus 50 in this study). All samples were immediately stored at 20 8C until analysis. Samples were given a unique sample code, the status of allergen labeling, a product- and subcategory code, and a unique lot number. The 899 samples consisted of individual lots of different brands belonging to 10 major product groups further divided into subcategories (supplementary information, Table A). From each of the 899 individual lots a minimum of three packages were collected to form at least 300 g of product. 633 (70%) samples were not particularly labeled for peanut and 266 samples (30%) were labeled for peanut traces. 2.2. Reference material Reference material (dark chocolate spiked with 0 and 2 ppm roasted peanut) was derived from the national research project ‘Development of rapid tests and screening methods for on-site detection of food allergens in product development and control’ (grant no. 132-281 6400508; German Ministry of Food and Agriculture, BMEL) and produced by IfP (Institute for Product Quality, Berlin, Germany). 2.3. Homogenization One third of each individual package of different weight was taken (e.g. 3 1/3 chocolate bars out of 10 chocolate bars in one package, or 170 g of a 500 g package of breakfast cereals) to yield 300 g and all subsamples per lot were pooled for homogenization. Dry and coarse samples (e.g. cereals, biscuits, appetizers) were crushed and homogenized by mixing for at least 2 min at 2000 rpm in a Grindomix GM300 laboratory mill (Retsch, Haan, Germany). More or less soft samples (e.g. cake) or those that were covered by ingredients of greasy texture (e.g. pizza) were first shock-frozen in liquid nitrogen and subsequently processed by milling. This procedure was also applied to chocolate bars (e.g. with caramel), other chocolate products and ice cream with tree nut particles. Chocolate tablets without visible particles, chocolate spread and ice cream were melted in a water bath (45 8C) and then homogenized by agitating. Cacao powder was mixed for at least two minutes in a three-dimensional shaker (TURBULA, Bachofen AG, Muttenz, Switzerland). 2 100 mL or g of each homogenate was filled into two separate sample vials. The A-sample was directly analyzed, while a B-sample was retained at 20 8C for repeated analysis or cross-check. 2.4. Lateral flow assay (LFA) All food samples were screened for peanut traces using a peanut lateral flow immunoassay kit (ImmunoFast, No. IF1002, IfP, Berlin) according to the manufacturer’s instructions. The limit of detection (LOD) of the test is specified as 2 ppm total peanut. The rapid stripe test includes an extraction buffer and is developed within 5 min. Incubation times were kept equal in all measurements. Two independent extractions from 200 mg or mL of homogenized sample were prepared. The optical density of the developed tests was monitored with a lateral flow immunoassay reader (OPTricon, Berlin, Germany) and calibrated against the reference materials. Samples yielding positive or suspect scores at least in one of two independent measurements were further investigated by real-time polymerase chain reaction (PCR). In addition, 5% of the double-negative samples in the LFA were randomly selected and also subjected to realtime PCR analysis.

198

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

2.5. Desoxyribonucleic acid (DNA) extraction Genomic DNA was isolated from 200 mg material of each sample by CTAB (cetyltrimethyl ammonium bromide) extraction according to the German official method BVL L-15.06.1. Briefly, 1 mL CTAB extraction buffer (CTAB 20 g/L, NaCl 1 M, Tris 0.1 M, Na2EDTA 20 mM) was added to the homogenized sample and the suspension was agitated for 30 min at 65 8C in a thermomixer. 15 mL of 20 mg/mL Proteinase K solution (Macherey-Nagel, Du¨ren, Germany) was added, the suspension incubated for 90 min at 65 8C and centrifuged (10 min, room temperature, 14,500  g). The supernatant was added to one volume ReadyRed (MP Biomedicals, Santa Ana, CA, USA) and mixed for 30 s. After centrifugation the clear supernatant was added to two volumes of CTAB precipitation buffer (CTAB 5 g/L, NaCl 0.04 M) and incubated for 60 min at room temperature. The precipitate was separated from the solution by centrifugation. After discarding the supernatant the pellet was solubilized in 300 mL NaCl (1.2 M) and incubated for 30 min at 37 8C. 10 mL of a 40 mg/mL RNAse solution (Macherey-Nagel) was pipetted to the solution and incubated for 10 min. 300 mL of ReadyRed was added to the solution, mixed for 30 s and centrifuged. The supernatant was mixed with 0.6 volumes of isopropanol and incubated overnight at 4 8C. The precipitated DNA was isolated by centrifugation (30 min, 4 8C, 14,500  g). The pellet was washed with 500 mL cold ethanol (70% v/v; 20 8C) and centrifuged (10 min, 4 8C, 14,500  g). Ethanol was discarded and the pellet was dried in a vacuum concentrator for 5 min. The dried DNA was dissolved in 50 mL water and stored at 20 8C until analysis. In case of fatty matrices such as chocolate, genomic DNA was isolated with the SureFood PREP Allergen kit (Congen, Berlin, Germany). The nucleic acid concentration and purity (absorption ratio at 260/280 nm) were measured with a NanoDrop UV-VIS spectrophotometer (Kisker, Steinfurt, Germany). Undiluted extracts were used for PCR. 2.6. Real time PCR Ready to use 96-well PCR microtiter plates were prepared as described previously (Zagon et al., 2012) and stored at room temperature until use in sealed aluminum bags with silica gel. Oligonucleotides and control DNA were pre-spotted and dried overnight on standard PCR microtiter plates together with trehalose in a mass ratio of approx. 20:1 nucleic acid:trehalose (supplementary information, Table B). Peanut DNA was detected according to Hird et al. (2003), but with deviating primer and probe concentrations of 300/300/200 nM primer forward/primer reverse/probe. The presence of amplifiable DNA or of PCR inhibitors in food extracts was examined with the universal plant DNA PCR system plant-nes according to Laube et al. (2010) with the same primer/probe concentration as for the peanut system. To check PCR inhibition, 20 copies of genomic hazelnut DNA were added into selected cavities in a dilution yielding a constant Ct-value of about 33 in PCR with water instead of sample DNA in plant-nes PCR. 5 mL of undiluted sample DNA or water in case of the nontemplate controls were pipetted to the mastermix (TaqMan Universal Master Mix, Applied Biosystems, Foster City, CA, USA). The standard time-/temperature program consisted of 45 cycles for 2 min at 50 8C, 10 min at 95 8C, 15 s at 95 8C, and 60 s at 60 8C. All measurements were run on a 7900 Fast Real-Time PCR-System (Applied Biosystems). For means of comparison the threshold for data evaluation of the PCR runs on four different days was kept constant at 0.2 for each run. The LOD of the peanut-specific system on the ready to use plates in terms of copy numbers was determined in a pre-validation with serial dilutions in 10 replicates. The LOD was defined as the lowest copy number at which all 10 replicates were still analyzed positive. This copy

number was 5 genome copies with an average Ct value of 41.67 and a standard deviation of 1.21. This Ct corresponded to 5 ppm total peanut in dark chocolate with a fixed DNA input of 100 ng/PCR, or to 2 ppm total peanut if measuring non-inhibited undiluted DNA extracts in PCR, respectively. 2.7. Enzyme-linked immunosorbent assay (ELISA) Positive samples in LFA and PCR were quantified for peanut traces using two different commercial ELISA kits (for details please refer to the supplementary information, Table C) on four days, namely ‘AgraQuant F.A.S.T. Peanut’ (Romer Labs, Tulln, Austria; hereafter termed ‘ELISA I’) and ‘RIDASCREEN FAST’ (R-Biopharm, Darmstadt, Germany; hereafter termed ‘ELISA II’). The kits were applied according to the manufacturer’s instruction on 1 g of homogenized sample. All washing steps were performed automatically with an ELISA well washer (Thermo Fisher Scientific, Waltham, MA, USA). 2.8. Calculation and statistics Unknown samples in quantitative ELISA were measured in triplicates, standard curve points as duplicates. The peanut concentration in ppm (mg/kg) was calculated automatically by SigmaPlot software. For both ELISA kits the best standard curve fit was observed with the nonlinear regression, polynomial ‘Cubic in Power-3-Para’ (f(x) = y0 + ax + bx2 + cx3) mode. A coefficient of determination R2 between 0.9998 and 1 and a standard error estimate between 0.0049 and 0.0742 was calculated for all standard curves. PCR data on unknown samples were gained from two independent extracts per sample and in double determination yielding four Ct-values per sample in two separate runs. The allover accordance of the obtained analytical results with the labeling status of the products was calculated by: ðPA þ NAÞ ð7 þ 631Þ ¼ ðPA þ ND þ PD þ NAÞ ð7 þ 259 þ 2 þ 631Þ The percentage of samples labeled for peanut traces and confirmed by analysis by: PA 7 ¼ ðPA þ NDÞ ð7 þ 259Þ The percentage of samples unlabeled for peanut traces and confirmed by analysis by: NA 631 ¼ ðPD þ NAÞ ð2 þ 631Þ whereas PA = number of positive agreements; NA = number of negative agreements; ND = number of negative deviation (false negatives); PD = number of positive deviation (false positives). To examine dependencies between labeling status, product categories and analytical results, Fisher’s 2-tailed exact test for independence was applied with a level of significance of a = 0.05 for rejecting the null hypothesis. 3. Results A schematic overview of the strategy of analysis and the results is presented in Fig. 1. 3.1. Lateral flow assay All samples were initially investigated by a rapid LFA screening test. The readout of this qualitative test was monitored in a dipstick reader, which was calibrated with dark chocolate reference

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

899 samples homogenized

760 negaves in double determinaon

899 samples LFA-screened (double extracon/sample)

5 % (41) randomly selected

89 ‘suspect‘ samples

50 posives (11 in double det.)

180 samples underwent confirmaon by real-me PCR (double extracon/sample, 2 technical replicates) posives by PCR and LFA, and samples with divergent LFA results

ELISA (30 samples)

199

signals between 1.4 and 3.5 million counts, which was in the range of the 2 ppm reference sample (data not shown). Altogether about 15.5% of samples were found positive or suspect in LFA prescreening, whereas about 84.5% of samples turned out to be negative in double determination. Altogether 50 ‘positive’ and 89 ‘suspect’ (S = 139) samples were further analyzed by real-time PCR. In addition, to confirm negative results, 41 samples (5%) from the remaining 760 double-‘negative’ samples were randomly selected for PCR investigation. 3.2. Real time PCR

9 posives detected and quanfied by ELISA

Fig. 1. Analytical workflow and summary of results. The graph shows a schematic representation of the analytical workflow consisting of a first screening for the presence of peanut by lateral flow assay (LFA), a second layer of analysis by realtime polymerase chain reaction (PCR), followed by quantification by commercially available enzyme-linked immunosorbent assay (ELISA) kits. The number of samples analyzed in each step is also given in the figure to summarize the results of the study.

samples containing 0 or 2 ppm of peanut. The reference materials were co-measured on each of 47 days of analysis. Over this period a range from 0 to 438,540 counts (mean 97,224; standard deviation (SD) = 137,964) was monitored for the negative sample and of 805,154–4,843,211 counts (mean 1,546,528; SD = 579,211) for the positive 2 ppm total peanut reference sample. Maximum values obtained for the 0 ppm sample and minimum scores for the 2 ppm sample did not overlap. To capture the range in between, a third category of ‘suspect’ results was defined empirically (Supplemental Figure 1). Thus, values 0.5 million counts were considered negative, between 0.5 and 0.97 million counts (mean of 2 ppm positive reference sample 1 SD) were classified as ‘suspect’, and values >0.97 million counts were categorized as positive results. Using this ‘three-level’ concept, a total of 50 samples out of 899 showed signals surpassing 0.97 million counts at least in one of the two LFA tests and were therefore classified ‘positive’. In addition, 89 samples were identified which yielded counts between 0.5 and 0.97 millions in at least one LFA test, while none of the two LFA tests produced >0.97 million counts with these samples. Thus, the 89 samples were classified ‘suspect’. The results of the LFA testing are summarized in Table 1. Out of the 50 positive samples, LFA signal strength pointed to a major peanut contamination in only three samples (no. 215 ‘pistachios’, no. 220 ‘aperifruits’ and no. 224 ‘cashew nuts’), which all belonged to the category ‘appetizer products’. All other 47 samples classified ‘positive’ in LFA screening yielded rather weak

Table 1 Summary of the results of lateral flow assay (LFA) screening of 899 food samples.

Out of the 139 ‘positive’ or ‘suspect’ samples investigated by real-time PCR, nine samples contained peanut DNA surpassing the LOD of the method. Most of these samples showed only weak or moderate fluorescence signals, but even in case of high Ct-values of 40 the obtained PCR signals were reproducible for both extracts in double determination as demonstrated by the low inter-assay coefficient of variation (Table 2) and absence of PCR inhibition (see below). Three samples (no. 181 ‘croustilles Emmental’, no. 193 ‘souffles fromages’, no. 219 ‘cashew nuts’) produced Ct-values about 40  2, which is in the range of the LOD of the method (approx. 5 genome copies). Five samples had Ct values between 32 and 38 (no. 215 ‘pistachios’, no. 220 ‘aperifruits’, no. 224 ‘cashew nuts’, no. 650 ‘milk chocolate with tree nuts’ and no. 750 ‘pate amande’). The highest signal was observed for a pistachio nut sample (no. 216 ‘pistachios’) with Ct values about 28–29, which points to a significant contamination. The remaining 130 samples were either unambiguously negative in all four PCR measurements per sample or, at the very most, characterized by a weak signal in only one out of four measurements. In the same way all 76 DNA extracts from samples double-negative in LFA testing were also negative in the PCR analysis (data not shown). In accordance with the LFA test results, positives were found among the ‘appetizer products’ group. Other samples (e.g. belonging to the ‘pizza’ and ‘cereal bar’ groups) which were ‘suspect’ in LFA screening could not be confirmed by PCR. All DNA extracts were of sufficient concentration and quality, as proven by the inhibition control with plant DNA (hazelnut) which was performed in double determination per sample (Table 2). Ct values surpassing a score of 33 in the presence of sample DNA would have indicated significant inhibition, which was not observed with any of the samples. Samples which were unequivocally positive or double-‘suspect’ in the LFA screening, as well as samples with strongly differing LFA outcome (one strongly positive result plus one negative or suspect result) were selected for further analysis by ELISA (Table 3). In total, a number of 30 samples were selected for quantification by ELISA, including the 9 LFA-positive samples which had been confirmed by PCR.

Results per sample (two independent extracts)

Overall scoring:

No. of samples

3.3. ELISA

Positive/Positive Positive/Suspect Positive/Negative

POSITIVE POSITIVE POSITIVE POSITIVE total: SUSPECT SUSPECT SUSPECT total: NEGATIVE NEGATIVE total:

11 4 35 50 8 81 89 760 760

ELISA data and a comparison of the results obtained with all applied methods are summarized in Table 3 for all thirty samples investigated (raw data including standard deviation see supplementary information, Table D). All nine samples which had been identified by LFA and subsequently confirmed by PCR revealed measureable peanut traces also in the ELISA test. Seven of those samples belonged to the ‘appetizer products’ group, while two samples, no. 650 and no. 750, belonged to ‘chocolate tables or chocolate spreads’. Samples no. 750 (almond spread) and no. 219 (nut/seed mixture) were not labeled for peanut traces and produced only weak signals only in ELISA I. In addition, sample no. 750 showed a considerable standard deviation between the triplicates underlining that the measurement reached the lower limits of the method.

Suspect/Suspect Negative/Suspect Negative/Negative

Total number of LFA-tested samples:

899

Total number of ‘POSITIVE’ + ‘SUSPECT’ samples investigated by real-time polymerase chain reaction (PCR):

139

Measured scores and interpretation: 500,000 counts: negative; 500,000–967,317 counts: suspect; >967,317 counts: positive.

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

200

Table 2 Positive samples in lateral flow assay (LFA) confirmed by real-time polymerase chain reaction (PCR). Sample no. Extract -1 and -2

Product category

Product

Labeled for peanut traces

Labeled for other allergen traces

DNA (mg/mL)

Purity (A260/280)

Cta Peanut

SD

COV (%)

Ct Inhibition control (Plant)b

181-1 181-2 193-1 193-2 215-1 215-2 216-1 216-2 219-1 219-2 220-1 220-2 224-1 224-2 650-1 650-2 750-1 750-2

Appetizer biscuits, cakes

Croustilles Emmental

Yes

Yes

Souffles fromages

Yes

Yes

Nuts, mixtures of seeds, fruits

Pistaches

Yes

Yes

Nuts, mixtures of seeds, fruits

Pistaches grillees et salees

Yes

Yes

Nuts, mixtures of seeds, fruits

Noix de cajou grille´es et sale´es

No

Yes

Nuts, mixtures of seeds, fruits

Aperifruits raisins, ananas, papayes, bananes, noix de coco Noix de cajou grille´es et sale´es

Yes

Yes

Yes

Yes

Chocolat lait raisins noisettes

Yes

Yes

Pate amande

No

Yes

1.67 1.66 1.72 1.77 1.81 1.76 1.63 1.61 1.35 1.47 1.76 1.10 1.51 1.7 1.64 1.69 1.98 1.99

38.46 40.18 39.15 39.22 36.21 36.01 28.91 28.44 42.92 40.97 34.31 32.15 34.78 33.24 37.74 36.98 38.08 37.88

0.74 0.18 0.08 0.05 0.15 0.33 0.29 0.47 0.47 0.13 0.33 0.21 0.37 0.28 0.22 0.00 0.18 0.21

1.17

Appetizer biscuits, cakes

53.61 44.91 52.86 43.58 125.78 152.37 275.01 301.70 161.81 138.63 44.70 55.10 123.60 137.20 28.70 53.60 260.00 265.60

16.08 15.67 15.02 15.34 18.99 18.74 16.65 16.04 20.12 27.66 18.04 18.56 26.44 23.76 22.38 21.79 16.71 16.54

Nuts, mixtures of seeds, fruits Milk chocolate bars with tree nuts Chocolate spread

0.17 0.66 1.33 0.72 0.81 0.96 0.59 0.51

Abbreviations: DNA, desoxyribonucleic acid; SD, standard deviation; COV, coefficient of variation. a Mean values of n = 2 replicates/extract. b Significant PCR inhibition would be indicated by Ct values 33.

An excellent correlation of ELISA I with the monitored PCR signal strength was observed. Only one sample, a biscuit (no. 308) with slightly elevated scores in the LFA test contained no peanut DNA detectable by PCR (Table 3). This sample was confirmed by ELISA to contain trace amounts of peanut above the LOD, but slightly below the limit of quantitation. The DNA extract of this sample was not inhibited (data not shown). Therefore neither an uneven distribution of the analyte nor a higher degree of DNA degradation can be excluded. ELISA method II produced, for most samples, results comparable ELISA method I. The great divergence among the two ELISA methods for sample no. 220, a mixture with fruits, could be due to inhomogeneity or matrix effects (e.g. adherence of the analyte to particles). This sample also gave markedly differing results with one and the same ELISA method on two different days (supplementary information, Table D). However, a tendency towards underestimation was observed using ELISA II for the reference sample as well as for some selected food samples, which is in line with the lower sensitivity claimed for this test. The correlation with PCR signal strength was generally weaker for ELISA II, as compared with ELISA I. 3.4. Statistical analysis The allover accordance of the analytical results with the labeling status was 70%. The percentage of samples which were labeled for peanut traces and confirmed by analysis was about 2.6%. The percentage of samples unlabeled for peanut traces and confirmed by analysis was 99.7%. The observed disagreements relate to the high number of products (97.4%) which were labeled for peanut traces, but did not contain measurable amounts above the LOD of the methods. To further investigate in how far the measured peanut contamination depended from the food category and labeling status, Fisher’s exact test for independence was applied. The results and conclusions are summarized in Table 4. For a detailed overview of the different food categories and sub-categories, please refer to the supplementary information, Table A. As detailed in Table 4, peanut contamination was strongly connected to the labeling status of the products. Statistical analyses also indicated a strong dependency of peanut contamination from the food categories: food products belonging to the category ‘appetizer

products’ contained traces of peanuts at significantly elevated frequency. The aforementioned category consisted of the two subgroups ‘appetizer biscuits, cakes’ and ‘nuts, mixtures of seeds, fruits’, the latter of which was contaminated with peanut at significantly higher frequency. From this it follows that the most critical group in this study was the ‘appetizer products’ category and among this the subcategory ‘nuts, mixtures of seeds, fruits’. 4. Discussion In order to investigate 899 samples from the French food market for the presence of peanut traces three different approaches were combined starting from an immunological LFA rapid test, followed by real-time PCR and finally by quantitative ELISA. Consequently, the reliability of the starting step is of utmost importance. The average values obtained for the reference materials in day-to-day LFA testing confirmed the specified LOD and functionality of the test with the peanut cultivar and processing degree of the reference material used in this study. Nevertheless a clear tendency for false positive results not confirmed by upstream analysis was observed for the LFA test. Elevated LFA scores occurred particularly with fatty and smeary matrices causing very faint or diffuse and doubtful bands which could be due to matrix effects rather than a specific test reaction. Basically, as a rule, false positive results in an entrance test are less problematic than a lack of sensitivity in this step. With a specified LOD of 2 ppm roasted total peanut, the LFA test used in this study reaches the upper sensitivity limit of a commercial immune stripe test for peanut detection known to the authors. Furthermore the rapid LFA test can be developed within five minutes and thus enabled convenient throughput of the huge sample number. In summary, the LFA screening yielded reliable results with the reference material and the procedure was easy and economical in terms of price and working time per sample. 4.1. PCR analysis It is to expect that the common food types as investigated in the frame of this project should contain, apart from peanut protein, also traces of peanut DNA. By using a PCR method in between the

Table 3 Comparison of LFA, PCR and ELISA results for 30 food samples. Food category

Original denomination

Labeled for peanut traces

Labeled for other allergen traces

LFA Extr. 1 Cts [Mio]

LFA Extr. 2 Cts [Mio]

Res.

PCR Peanut Cta

ELISA I ppmb Total peanut

ELISA II ppmb Total peanut

26 50 111 115 116 130 152 166 181 193 215 216 219 220

Breakfast cereals with chocolate Cereal bars with chocolate Viennoiseries with chocolate Viennoiseries with chocolate Viennoiseries with chocolate Viennoiseries with chocolate Viennoiseries plain Pitas, tortillas, crackers Appetizer biscuits, cakes Appetizer biscuits, cakes Nuts, mixtures of seeds, fruits Nuts, mixtures of seeds, fruits Nuts, mixtures of seeds, fruits Nuts, mixtures of seeds, fruits

No Yes Yes No Yes Yes No No Yes Yes Yes Yes No Yes

Yes Yes Yes Yes Yes Yes No No Yes Yes Yes Yes Yes Yes

1.06 0 0 0.62 0.67 0.77 0.14 1.30 1.98 1.66 5.18 0.69 1.19 10.47

1.64 1.89 1.48 0.82 1.43 0.94 0 1.83 1.82 3.54 3.36 1.23 1.21 6.29

POS POS POS SUS POS SUS SUS POS POS POS POS SUS SUS POS

neg. neg. neg. neg. neg. neg. neg. neg. 39.32 W 0.46 39.19 W 0.07 36.11 W 0.24 28.68 W 0.38 41.95 W 0.3 33.23 W 0.27


n.i. n.i. n.i. n.i. n.i. n.i. n.i. n.i. 3.9 W 0.6 3.6 W 0.7 2.3 W 0.4 10 W 1.5
224 241

Nuts, mixtures of seeds, fruits Cream desserts, mousse or fresh desserts with chocolate Cream desserts, mousse or fresh desserts with chocolate Biscuits and pastry with chocolate Biscuits and pastry with chocolate Biscuits and pastry with chocolate Biscuits and pastry with fruits, dried fruit Milk chocolate bars with tree nuts White chocolate bars Chocolate spread Chocolate spread Chocolate spread Bonbons with chocolate Icecream with chocolate Icecream with fruits, dried fruits Icecream, other parfum Reference material Reference material

Tresor chocolat noisette Nesquick cereales et lait Brioches gout chocolat Brioches gout chocolat Brioches pepites de chocolat Pains chocolat au levain Briochettes levains Tortillas a garnir Croustilles Emmental Souffles fromage Pistaches Pistaches grille´es et sale´es Noix de cajou grille´es et sale´es Aperifruits raisins, ananas, papayes, bananes, noix de coco Noix de cajou grille´es et sale´es Fondant chocolat

Yes No

Yes Yes

6.09 0.38

3.43 1.47

POS POS

34.01 W 0.33 neg.

8 W n.c.
1.8 W 0.1 n.i.

Mousse chocolat

No

Yes

1.57

1.12

POS

neg.


n.i.

Gouters saveur chocolat Mini Gouters saveur chocolat Brownie chocolat pepites La Barquette trois chatons fraise Chocolat lait raisins noisettes Confiserie blanche Pate amande Nutella Nutella Petits oursons guimauve chocolat noir Smarties mini Sorbet framboise Disney Parade pouss pouss vanille Dark chocolate 0 ppm peanut Dark chocolate 2 ppm peanut

Yes Yes Yes No Yes No No No No No No Yes Yes – –

Yes Yes Yes Yes Yes Yes Yes No No Yes No Yes Yes – –

0 2.18 0 0.40 2.41 0.03 0.68 0.62 1.35 1.77 3.74 0.18 0.67

5.23 3.55 3.19 1.37 1.63 1.44 0.59 2.04 0.09 0 0.07 1.62 0.70 0–0.44 0.81–4.84

POS POS POS SUS POS POS SUS POS SUS POS POS POS SUS NEG POS

neg. neg. neg. neg. 37.36 W 0.22 neg. 37.98 W 0.2 neg. neg. neg. neg. neg. neg. neg. 42.36 W 1.5


n.i. n.i. n.i. n.i.
255 301 308 316 517 650 738 750 759 772 805 864 868 885

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

No.

Abbreviations: Cts, counts; Extr., extract; LFA, lateral flow assay; PCR, polymerase chain reaction; ELISA, enzyme-linked immunosorbent assay; POS, positive in LFA; SUS, suspect in LFA; NEG, negative in LFA; neg., no signal detected in PCR; n.i., not investigated; n.c., not calculable (<0.1); LOD, limit of detection. Please refer to Table C in the supplementary material for information about the detection limits of the ELISA kits. a Mean  SD from four measurements and two runs is shown. b Mean  3 SD from triplicate measurements is shown.

201

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

202

Table 4 Statistical analysis of results, as performed using Fisher’s exact test for selected groups. Investigated groups

Peanut contamination (Results of analysis)

Calculated P-value (2-tailed)a

Conclusion

Interpretation

Positive samples No./(%)

Negative samples No./(%)

Total investigated samples labeled for peanut traces (S = 266) Total investigated samples unlabeled for peanut traces (S = 633)

7 (2.6%) 2 (0.3%)

259 (97.3%) 631 (99.7%)

0.0038

Strong dependency of the result from the labeling status

Most positives found in ‘labeled’ group

Category ‘appetizer products’ (S = 58) Category ‘chocolate tablets and spreads’ (S = 140)

7 (12%) 2 (1.4%)

51 (88%) 138 (98.6%)

0.003

Strong dependency of the result from the category

Most positives found in ‘appetizer products’ group.

Sub category ‘appetizer biscuits, cakes’ (S = 44) Sub-category ‘nuts mixtures of seeds, fruits’ (S = 14)

2 (4.5%) 5 (35.7%)

42 (95.5%) 9 (64.3%)

0.007

Strong dependency of the result from the subcategory

Most positives found in the sub category ‘nuts, mixtures of seeds, fruits’.

Labeled ‘appetizer products’ (S = 40) Labeled ‘chocolate tablets or spreads’ (S = 8)

6 (15%) 1 (12.5%)

34 (85%) 7 (87.5%)

1.0000

No significant difference between labeled products in both groups and results of analysis

Percentage of positive and negative results in a comparable range in both groups

Unlabeled ‘appetizer products’ (S = 18) Unlabeled ‘chocolate tablets or spreads’ (S = 132)

1 (5.6%) 1 (0.8%)

17 (94.4%) 131 (99.2%)

0.2253

No significant difference between unlabeled products in both groups and results of analysis

Percentage of positive and negative results roughly in the same range in both groups

a

Level of significance a = 0.05.

LFA and ELISA testing, the certainty to catch true positive results is considerably enhanced since false positive results due to unspecific antibody cross reactions are excluded. Although the ELISA test kits have been specificity-tested by the supplier with a wide range of species, an antibody is a biomolecule which might exhibit unpredictable binding properties towards target molecules others than those empirically tested. In this sense, and against the background of food complexity, PCR is an ideal complementary method, since species specificity is strictly defined by a unique DNA sequence. For this reason PCR makes also part of the official allergen control for confirmation of ELISA results in Japan (Akiyama et al., 2011). However, in Japan a 10 ppm threshold for labeling has been implemented. The challenge in this study was to find a PCR system that would meet a sensitivity of at least 2 ppm as defined by the pre-screening test. Among six known real-time PCR systems targeting peanut DNA, three methods are reported to reach a LOD in a matrix of 10 ppm peanut (Stephan and Vieths, 2004; Scaravelli et al., 2008; Lopez-Calleja et al., 2013) and two methods are described to detect 2 ppm (Hird et al., 2003) and 0.1 ppm (Lopez-Calleja et al., 2013) in food, respectively. The latter rather recently published system targets the Internal Transcribed Spacer (ITS1) region of ribosomal RNA-coding DNA which is a highly repetitive genetic element. Despite being the most sensitive approach, this method was not considered because of the danger of cross contamination in routine applications. Therefore the PCR system according to Hird et al. (2003) was preferred which showed satisfying and stable performance during in-house validation. Most favorably, the targeted sequence of the Ara h 2 allergen gene has the shortest length (66 bp) among all published peanut PCR systems which makes it well suited for the detection of processed DNA. The method is reported to detect 2 ppm lightly roasted peanut in biscuits treated for 15 min at 160 8C. This sensitivity was confirmed with chocolate reference material spiked with 2 ppm roasted peanut. However, the fluorescence signal for this material was weak (>Ct 40) with 100 ng DNA input in PCR and only 13 out of 40 measurements yielded measurable signals (in average

42.36  1.5) in a homogeneity test in an in-house pre-validation. Because of this observation it was necessary to apply undiluted DNA to entrap even minute traces of peanut DNA and to reach the desired sensitivity of 2 ppm. In fact some samples in this study with high Ctvalues about 40 would have been overseen with diluted DNA in PCR. In our study 5 mL of sample DNA was added to the mastermix, corresponding to total DNA amounts between 145 ng at minimum and 1375 ng at maximum. Therefore the aspect of inhibitor-free purity of the extract was essential. In this respect the CTAB protocol used in this study delivered DNA of high quality as proven by the inhibition control experiments. However, for certain chocolate formulas a commercial kit had to be used to yield satisfying results. Using well-purified extracts the PCR signals correlated well with ELISA method I, whereas a strict relationship between the signal strength in LFA and PCR was not obvious. Only one biscuit sample with chocolate (no. 308) revealed ELISA signals with one test kit slightly over the LOD but was not negative in PCR testing. Though an exceptional finding, this result demonstrates the usefulness to rely not only on one test principle. The low standard deviations of Ct-values obtained on different days underline the excellent repeatability and reproducibility of the ‘ready to use’ plates used in this study. The positive impact of reagent pre-spotting on the reproducibility of results from different PCR runs has been shown previously in a validation study on the method (Zagon et al., 2012). Since all primers and probes are pre-pipetted, the operator just has to apply DNA extract and mastermix into the cavities of the microtiter plates. This minimizes pipetting errors. Moreover, a PCR-enhancing effect might be assumed due to the presence of the bioprotective agent trehalose (Spiess et al., 2004). 4.2. ELISA analysis Two different test kits were applied in the final ELISA quantification. The test kit termed ELISA I (AgraQuant F.A.S.T. Peanut, Romer Labs) in this investigation belongs to the most

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

sensitive commercially available systems with a LOD of 0.5 ppm total peanut. The alternative system ELISA II (RIDASCREEN FAST, RBiopharm) is less sensitive with a specified LOD of 1.5 ppm total peanut, but was successfully performance-tested with 5 ppm incurred material in a ring trial (AOAC-RI-030404) with three laboratories (Park et al., 2005). ELISA II is reported to cross react with green pea, lima beans and chick pea, but these species were not of relevance to the matrices under investigation in our study. The results showed that ELISA I indeed achieved a better sensitivity than ELISA II which is in line with the specified LODs of the methods. Peanut traces above 5 ppm were detected without problems by both kits. Observed signal differences between both tests might also be a matter of analyte distribution. Although thoroughly homogenized, neither ‘hot spots’ can be completely ruled out in case of differently sized micro particles, nor can a distribution effect for contaminations be excluded at the LOD of the test methods. In this context the number of observations per sample in this study is too low for an exact comparison between both ELISA kits. In addition it is well known that the performance of an ELISA test also depends on the matrix under investigation (AOAC, 2012). Therefore absolute ppm values obtained by ELISA tests with various matrices others than those taken in the validation study must be interpreted with caution. In the light of measurable peanut traces the labeling status of the products is of great concern. Seven of nine positive samples with peanut traces belonged to the ‘appetizer products’ group and from these, five samples to the sub-category ‘nuts, mixtures of seeds, fruits’ which clearly underlines the high relevance of this group to contain a contamination. Six out of the seven positive ‘appetizer products’ samples were correctly labeled for the presence of peanut traces including the sample with the highest measurable content of up to 17 ppm (ELISA I) total peanut. Only one cashew nut sample was not declared for peanut but contained weak traces in the range of approximately 1.5 ppm total peanut in ELISA I, which was below the LOD of ELISA II. From the two positive samples of the category ‘chocolate tablets or chocolate spread’ a milk chocolate product with tree nuts containing approximately 3 ppm total peanut in ELISA I (negative in ELISA II), was correctly labeled for peanut traces. The second positive sample was not labeled for peanut but contained only faint traces (0.7 ppm total peanut) near the LOD of the highly sensitive ELISA I method. The measured contamination levels in this study are in accordance with previously published investigations on products of these categories (Elegbede et al., 2015). In total, the analytical results confirmed that 99.7% of all unlabeled products (633) were correctly declared negative. On the other hand, only a minor proportion of 2.6% of all positive-labeled products (266) did contain measureable peanut traces above 2 ppm total peanut with the method cascade used in this study. This observation underlines the high incidence of precautionary labeling of products on the market. 4.3. Peanut contamination and human risk The choice of categories and amount of products per category and sub-category was defined by an optimized sampling plan based on a Bayesian network considering consumption probabilities of consumers (including children’s habits), the knowledge on unintended presence according to literature and the frequency of precautionary labeling (Elegbede et al., 2015). According to this, labeled ‘chocolate tablets or chocolate spreads’ should have the highest probability of peanut traces (40%), whereas ‘appetizer products’ the lowest (2.7%). However, no statistically significant difference could be found between these groups for labeled products in this study. In addition, the high level of peanut contamination in a range of 60 ppm reported for products

203

belonging to ‘breakfast Cereals’, ‘cereal bars’ and ‘other chocolate products’ in a previous study (Elegbede et al., 2015) was not confirmed. It has to be mentioned that the underlying literature on the status of peanut contamination was published between 1999 and 2007. Furthermore the number of investigated samples by Elegbede et al. (2015) was lower, as compared to the present study. Improved quality control and hazard concepts by manufacturers as well as the sample size chosen in this study might be an explanation for the deviating results. The LOD of the methods was 2 ppm total peanut. Therefore the question arises whether exposure to contaminations with 2 ppm peanut (0.5 mg peanut protein/kg) will pose a risk to allergic consumers. In fact it is known that already small amounts of peanut in the lower ppm range may trigger allergic symptoms (Hourihane et al., 1997; Taylor et al., 2002). The study of Wensing et al. (2002) who performed a double-blind placebo-controlled food challenge test with 26 peanut allergic patients revealed a threshold dose between 100 mg and 1 g peanut protein, with 30 mg being the lowest dose eliciting a response in this study. Therefore, the consumption of 60 g of a material like chocolate or biscuits contaminated with 2 ppm (2 mg/kg) whole peanut, corresponding to approximately 0.5 ppm pure peanut protein (25% of total peanut w/w), would be sufficient to provoke a reaction in the most sensitive persons. The long-term study by Taylor et al. (2010) on a bigger cohort of 286 peanut allergic persons resulted in a higher LOAEL (lowest observed adverse effect level) between 0.5 mg and up to 8–10 mg total peanut (approx. 125–2500 mg peanut protein). Hence the critical amount would be achieved after consumption of 500 g of food containing 2 ppm whole peanut. These examples, disclosing widely variable LOAELs, underline the difficulty to compare the results of different clinical studies and to infer general conclusions. Furthermore, the severity of symptoms largely depends on the individual condition and is hardly reproducible (Glaumann et al., 2013). Therefore attempts have been made to define the biggest possible amount of peanut that can be consumed by the total population without adverse effects by probabilistic statistical models. Taylor et al. (2009, 2010) deduced an ED10 (effective dose; the total peanut amount which is necessary to provoke a reaction in 10% of the peanut allergic population) between 8.4 and 14.4 mg, and an ED05 of 7.3 mg. According to the study of Rimbaud et al. (2010) on the threshold dose eliciting adverse reactions, the ED05 is at 1.2 mg of peanut protein and the ED01 is at 0.05 mg. At least 99% of peanut allergic persons will react to a dose 6.4 mg of peanut protein. 4.4. Summary and conclusion As integral part of the MIRABEL project, this study was aimed to analyze the real market situation regarding the presence and amount of unintended peanut traces and to assess and quantify the all over risk against the background of analytical results, medical data and individual consumption data. To the author’s knowledge a combination of immunological and DNA-based methods with a limit of detection of 2 ppm total peanut was applied for the first time to investigate a broad range of everyday products, thus approaching the boundaries of sensitivity in current peanut detection systems. In our study only three out of 899 samples surpassed a margin of 8 mg/kg whole peanut, corresponding to 2 ppm peanut protein and all of them were correctly labeled. This indicates a satisfying allergen management in terms of peanut traces. Nevertheless, the existence of ‘hot spots’, i.e. unevenly distributed single particles of an allergenic compound, can never be ruled out. The integrated MIRABEL approach which was performed for the French population is also indicative for other European countries. The results are expected to be of great interest for risk assessment

204

J. Zagon et al. / Journal of Food Composition and Analysis 44 (2015) 196–204

and regulatory bodies as well as for food industry. Furthermore, it is to expect that the project’s output will support the pending discussion on the introduction of thresholds in Europe. However, any decision on future thresholds must take into account the balance between medical implications and health risks, practical feasibility by producers, and last but not least the possible surveillance by food control bodies. Our results on hundreds of shelf samples revealed that the detection down to 2 ppm of total peanut in different kinds of products is feasible with a spectrum of methods as used here. Acknowledgements This project was funded in the frame of the project MIRABEL which is supported by the French Research Agency (ANR) under the reference ANR-10-ALIA-2012. The authors thank Annika Franke, Stefanie Gegner, Melanie Kotulla, Bettina Linke, Antina Ogungbemi, Beatrice Rosskopp and Anja Witte for excellent assistance. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jfca.2015.08.006. References Akiyama, H., Imai, T., Ebisawa, M., 2011. Japan food allergen labeling regulation – history and evaluation. Adv. Food Nutr. Res. 62, 139–171. AOAC, 2012. Official Methods of Analysis. Food Allergen Community Guidance. Appendix M. Retrieved May 29, 2015 from: http://www.eoma.aoac.org/ app_m.pdf. Cre´pet, A., Papadopoulos, A., Elegbede, C.F., Loynet, C., Ait-Dahmane, S., Millet, G., van der Brempt, X., Bruye`re, O., Marette, S., Moneret-Vautrin, D.A., 2015. MIRABEL: an integrated framework for risk and cost/benefit analysis of peanut allergen. Regul. Toxicol. Pharmacol. 71, 178–183. EFSA, 2004. Opinion of the Scientific Panel on Dietetic Products, Nutrition and Allergies on a request from the commission relating to the evaluation of allergenic foods for labelling purposes (Request No EFFSA-Q-2003-016). EFSA J. 32, 1–197. Elegbede, C.F., Papadopoulos, A., Gauvreau, J., Cre´pet, A., 2015. A Bayesian network to optimise sample size for food allergen monitoring. Food Control 47, 212–220. Glaumann, S., Nopp, A., Johansson, S.G.O., Borres, M.P., Nilsson, C., 2013. Oral peanut challenge identifies an allergy but the peanut allergen threshold sensitivity is not reproducible. PLoS One 8, e53465. Grundy, J., Matthews, S., Bateman, B., Dean, T., Arshad, S.H., 2002. Rising prevalence of allergy to peanut in children: data from 2 sequential cohorts. J. Allergy Clin. Immunol. 110, 784–789. Hourihane, J.O., Kilburn, S.A., Nordlee, J.A., Hefle, S.L., Taylor, S.L., Warner, J.O., 1997. An evaluation of the sensitivity of subjects with peanut allergy to very low doses of peanut protein: a randomized, double-blind, placebo-controlled food challenge study. J. Allergy Clin. Immunol. 100, 596–600.

Hourihane, J.O., 2011. Peanut allergy. Pediatr. Clin. North Am. 58, 445–458. Hird, H., Lloyd, J., Goodier, R., Brown, J., Reece, P., 2003. Detection of peanut using real-time polymerase chain reaction. Eur. Food Res. Technol. 217, 265–268. Husain, Z., Schwartz, R.A., 2012. Peanut allergy – an increasingly common life threatening disorder. J. Am. Acad. Dermatol. 66, 136–143. Laube, I., Hird, H., Brodmann, P., Ullmann, S., Schone-Michling, M., Chisholm, J., Broll, H., 2010. Development of primer and probe sets for the detection of plant species in honey. Food Chem. 118, 979–986. Lopez-Calleja, I.M., de la Cruz, S., Pegels, N., Gonzalez, I., Garcia, T., Martin, R., 2013. Development of real-time PCR assay for detection of allergenic trace amounts of peanut (Arachis hypogaea) in processed foods. Food Control 30 (2), 480–490. Morisset, M., Moneret-Vautrin, D.A., Kanny, G., 2005. Prevalence of peanut sensitization in a population of 4,737 subjects – an Allergo-Vigilance Network enquiry carried out in 2002. Eur. Ann. Allergy Clin. Immunol. 37, 54–57. Moneret-Vautrin, D.A., 2008. E´pide´miologie de l’allergie alimentaire. Rev. Fr. Allergol. Immunol. Clin. 48, 171–178. Park, D.L., Coates, S., Brewer, V.A., Garber, E.A.E., Abouzied, M., Johnson, K., Ritter, B., McKenzie, D., 2005. Performance tested method multiple laboratory validation study of Elisa-based assays for the detection of peanuts in food. J. AOAC Int. 88, 156–160. Rance´, F., Grandmottet, X., Grandjean, H., 2005. Prevalence and main characteristics of schoolchildren diagnosed with food allergies in France. Clin. Exp. Allergy 35, 167–172. Rimbaud, L., Heraud, F., La Vieille, S., Leblanc, J.C., Cre´pet, A., 2010. Quantitative risk assessment relating adventitious presence of allergens in food: a probabilistic model applied to peanut in chocolate. Risk Anal. 30, 7–19. Sampson, H.A., Mendelson, L., Rosen, J.P., 1992. Fatal and near-fatal anaphylactic reactions to food in children and adolescents. N. Engl. J. Med. 327, 380–384. Scaravelli, E., Brohee, M., Marchelli, R., Hengel, A.J., 2008. Development of three real-time PCR assays to detect peanut allergen residue in processed food products. Eur. Food Res. Technol. 227, 857–869. Spiess, A.N., Mueller, N., Ivell, R., 2004. Trehalose is a potent PCR enhancer: lowering of DNA melting temperature and thermal stabilization of Taq polymerase by the disaccharide trehalose. Clin. Chem. 50, 1256–1259. Stephan, O., Vieths, S., 2004. Development of a real-time PCR and a sandwich ELISA for detection of potentially allergenic trace amounts of peanut (Arachis hypogaea) in processed foods. J. Agric. Food Chem. 52, 3754–3760. Taylor, S.L., Hefle, S.L., Bindslev-Jensen, C., Bock, S., Burks, A.W., Christie, L., Hill, D.J., Moneret-Vautrin, D.A., Vadas, P.A., Rance, F., Scrypec, D.J., Trautman, T.A., Malmheden, Y.I., Zeiger, R.S., 2002. Factors affecting the determination of threshold doses for allergenic foods: how much is too much? J. Allergy Clin. Immunol. 109, 24–30. Taylor, S.L., Crevel, R.W.R., Sheffield, D., Kabourek, J., Baumert, J., 2009. Threshold dose for peanut: risk characterization based upon published results from challenges of peanut-allergic individuals. Food Chem. Toxicol. 47, 1198– 1204. Taylor, S.L., Moneret-Vautrin, D.A., Crevel, R.W.R., Sheffield, D., Morisset, M., Dumont, P., Remington, B.C., Baumert, J.L., 2010. Threshold dose for peanut: risk characterization based upon diagnostic oral challenge of a series of 286 peanut-allergic individuals. Food Chem. Toxicol. 48, 814–819. Wensing, M., Penninks, A.H., Hefle, S.L., Koppelmann, S.J., Bruunzeel-Koomen, A., Knulst, A.C., 2002. The distribution of individual threshold doses eliciting allergic reactions in a population with peanut allergy. J. Allergy Clin. Immunol. 110, 915–920. Yunginger, J.W., Sweeney, K.G., Sturner, W.Q., Giannandrea, L.A., Teigland, J.D., Bray, M., Benson, P.A., York, J.A., Biedrzycki, L., Squillaceet, D.L., Helm, R.M., 1988. Fatal food-induced anaphylaxis. J. Am. Med. Assoc. 260, 1450–1452. Zagon, J., Kurth, S., Ehlers, A., Linke, B., Lampen, A., Broll, H., 2012. Preservation of primer and probes on ‘‘ready-to-use’’ 96-well micro titer plates: a step forward towards enhancing throughput and harmonization of real time PCR applications in food and feed control. Food Control 25, 709–716.