Evaluating key-processing parameters for reliable cross-sample analysis in comprehensive two-dimensional gas chromatography: exploring the human saliva metabolome in diet intervention studies

Comprehensive two dimensional gas chromatography with time of flight mass spectrometry (GCxGCTOF MS) represents one of the most powerful analytical platform for chemical investigation of complex samples. However, it produces large and complex sets of data, rich of information, but whose consistency might be affected by random fluctuations of system performances or changes in experimental parameters. This study focuses on human saliva metabolites signatures explored by GCxGC-TOF MS followed by untargeted and targeted (UT) pattern recognition, i.e., UT fingerprinting. Key-process parameters are examined for their impact on false negative matches and for consistent cross-alignment of data. Signal-to-noise ratio detection and MS spectrum similarity thresholds were systematically varied to generate reference patterns (i.e., templates) to be used for effective cross-alignment. To compensate for retention time misalignment, supervised procedures accompanied by global polynomial 2nd order transforms were tested. Case Study-I refers to a diet intervention by meals rich in advanced glycation end products (AGEs). The UT fingerprinting was applied to identify markers arising from a AGEs rich diet vs. a control diet. Case Study-II deals with metabolically healthy (MHO) and unhealthy (MUHO) obesity and saliva signatures were captured by TOF MS acquiring in Tandem Ionization (TI) conditions. The two datasets showed marked, random pattern shifts. By combining S/N and MS similarity thresholds to global polynomial 2nd order transforms and supervised realignment of patterns, the matching rate of reference 2D peaks increased from 51% to 84%. Once re-aligned, peak and peak-region features were explored by supervised pattern recognition to reveal potential markers.