Abstract
An authentication study based on headspace solid-phase microextraction/GC-MS was performed with a set of 60 samples representative of traditional “Pasta di Gragnano protected geographical indication (PGI)” and the most common Italian pasta brands. Multivariate chemometric tools were used to classify the samples based on the chemical information provided from 20 target flavor compounds, including Maillard reaction and lipid oxidation products. Pattern recognition by principal component analysis and linear discriminant analysis showed a natural grouping of samples according to the drying process adopted for their production (i.e., the traditional Cirillo method versus a high-temperature approach). Subsequently, soft independent modeling by class analogy (SIMCA) and unequal dispersed classes (UNEQ) were used to build class models at 95% confidence and 100% sensitivity levels (forced models) for predictive classification purposes. The good performance obtained from the models in terms of cross-validation efficiency (SIMCA, 57.01%; UNEQ, 86.60%; 100% for both forced models) highlighted that targeted analysis of flavor profiles could be used to assess the authenticity of Pasta di Gragnano PGI samples. Hence, the proposed method may help to protect Pasta di Gragnano PGI from label frauds by verifying whether samples comply with statements concerning drying process conditions as stated in the product specification.
Correction to: Identification of iron ore brands by multi-component analysis and chemometric tools
