In-house validation of a prototype reference method for six allergens detection in chocolate by HPLC-MS/MS analysis

In the last ten years, mass spectrometry has been increasingly exploited to detect allergens in food thanks to the capacity of techniques for multiplexing and providing unequivocal allergen identification. This methodology offers great potentials for designing a reference method for multiple allergen analysis in food commodities that is the major goal of the European ThRAll (Thresholds and Reference method for Allergen detection) project funded by the European Food Safety Authority. The method under development refers to a prototype quantitative method for multiple detection of food allergens in incurred food matrices and target six allergenic ingredients, including cow's milk, hen's egg, peanut, soybean, hazelnut and almond to be detected in two hard-to-analyse food matrices, namely chocolate bar and broth powder [1]. Both matrices were produced in a food pilot plant, to mimic as closely as possible the actual manufacturing processes and the allergenic ingredients were incurred in order to simulate the accidental contamination of matrices along the production line. Such materials were further used for the development of a multiplex HPLC-MS/MS method monitoring highly selective and reliable peptide markers reporting for contamination from the selected allergens [2, 3].In this communication, the advances obtained with the first incurred matrix produced, i.e. chocolate bar, will be presented. The prototype reference method was designed for absolute quantitation of allergenic ingredients by matrix matched calibration curves prepared with synthetic peptides and isotopically labelled surrogates as internal standards. The method was validated in-house by testing the main analytical performances such as linearity, sensitivity, matrix effect, specificity, precision and trueness. The response linearity was tested over two orders of magnitude (0.5-50 fmol/µL) for all the six allergens. The method provided different sensitivity depending on the specific allergen, and grounded on the detection of at least two peptide markers per allergenic ingredient at the lowest detected point except for egg white. Different analytical approaches available for the experimental calculation of detection and quantification limits (LOD/LOQ) were applied and discussed. For most of the reporting markers, the detection sensitivity resulted matrix-dependent by comparison of matrix-matched and standard calibration curves. The detection specificity was proved (i) in-silico by BLAST (Basic Local Alignment Search Tool) search of the peptide markers sequence against UniProtKB database and (ii) in-vitro by analysing blank chocolate bar. The method reproducibility (intra-day variation) and intermediate precision (inter-day variation) were evaluated on incurred samples at two concentration levels (4 and 40 ppm), resulting in CV% always below 20%. As for method trueness, due to the limited availability of proper reference materials (RMs) a dual approach was followed including (i) recovery experiments on spiked matrices and (ii) direct analysis of the only chocolate based RM currently available on the market and commercialized by LGC Standards. Finally, the method was tested by analysing incurred samples at decreasing concentration levels (40, 10, 4 and 2 ppm) proving its applicability at the lowest recommended VITAL doses for all the six allergenic ingredients. This method could represent, in perspective, the first quantitative reference method for egg, milk, soybean, peanut hazelnut and almond detection in chocolate based matrixes as result of ThRAll project.