Collaborative validation of the quantification method for suspected allergens and test of an automated data treatment

Previous publications investigated different data treatment strategies for quantification of volatile suspected allergens by GC/MS. This publication presents the validation results obtained on “ready to inject” samples under reproducibility conditions following inter-laboratory ring-testing. The approach is based on the monitoring of three selected ions per analyte using two different GC capillary columns. To aid the analysts a decisional tree is used for guidance during the interpretation of the analytical results. The method is evaluated using a fragrance oil concentrate spiked with all suspected allergens to mimic the difficulty of a real sample extract or perfume oil. At the concentrations of 10 and 100 mg/kg, imposed by Directive 76/768/EEC for labeling of leave-on and rinse-off cosmetics, the mean bias is +14% and −4%, respectively. The method is linear for all analytes, and the prediction intervals for each analyte have been determined. To speed up the analyst's task, an automated data treatment is also proposed. The method mean bias is slightly shifted towards negative values, but the method prediction intervals are close to that resulting from the decisional tree.

► The GC/MS quantitation of suspected allergens is validated thanks to ring test.
► The results are treated using a decisional tree or an automated data treatment.
► Both methods are linear, and the prediction intervals are given.
► From 10 to 100 mg/kg, the mean bias of both methods is below 15%.
► The automated treatment performances are close to those of the decisional-tree.