An effective chromatographic fingerprinting approach based on image and peak-region features generated by comprehensive two-dimensional gas chromatography - mass spectrometry: food quality applications

Comprehensive two-dimensional gas chromatography coupled with mass spectrometry (GC×GC-MS) is nowadays the most informative analytical approach for the chemical characterization of food volatilome. Key features of this technique are separation power and resolution enhancement, improved sensitivity and generation of structured separation patterns of chemically related groups of analytes. This contribution presents a workflow that combines image and peak-region features for effective chromatographic fingerprinting of complex patterns. Applications deal with food volatiles and their capability to identify diagnostic 2D patterns in spoiled hazelnuts classified by sensory flash-profiling (FP), and in extra-virgin olive oils (EVOOs) produced in Italy (Sicily, Tuscany and Garda area). The workflow includes the generation of composite 2D chromatograms, from samples belonging to the same class (i.e., cumulative class-images), that facilitate the identification of characteristic and distinctive patterns by pair-wise comparative analysis. Then, by exploring the distribution of untargeted/targeted peak-regions comprehensively mapped on the chromatographic space, potential markers can be selected and their information role validated. Results are encouraging and, for spoiled hazelnuts (Mould, Mould-rancid-solvent, Rancid, Rancid-stale, Rancid-solvent, and Uncoded spoilage), indicated that octanoic, heptanoic and hexanoic acids together with ƴ-octalactone, ƴ-nonalactone, ƴ-hexalactone, acetone, and 1- nonanol are decisive to discriminate quality hazelnuts from spoiled ones. For EVOOs, the lipoxygenase degradation products signature, the C7-C12 saturated and unsaturated aldehydes series, several terpenoids and a few known potent odorants are differentially distributed in samples from different locations. The workflow enables effective isolation of confounding variables and prompt identification (even visual) of diagnostic patterns of chemicals. Its effectiveness is proved for both thermal modulated (loop-type modulator) and differential-flow modulated (reverse-inject differential flow microfluidic device) GC×GC-MS(/FID) methods.