Extra Virgin Olive oil volatiles a mine of chemical information: challenges in chromatographic data alignment and response normalization for reliable fingerprinting by comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry.

Introduction Comprehensive two-dimensional gas chromatography combined with time-of-flight mass spectrometry (GC×GC-MS) is a highly informative fingerprinting technique for characterization of complex fractions of volatiles (VOCs) in food. In this study, 2D patterns of volatiles from Extra Virgin Olive oils (EVO oils), analyzed over a one-year time-frame by applying chromatographic set-up changes, are evaluated to compensate for 2D peak pattern shifts and response fluctuations to make consistent and reliable the Untargeted/Targeted (UT) fingerprinting process. Methods VOCs from high-quality EVO oils (Italian Violin Project) are sampled from samples Head Space (HS) by Solid Phase Micro Extraction (HS-SPME), using a DVB/CAR/PDMS df 50/30 μm 1 cm length fiber and analyzed by GC×GC-ToFMS adopting a polar × medium polarity set-up. By altering some chromatographic parameters known to affect 2D pattern characteristics (i.e., modulation period MP, 2D column dimensions and tuning of TOF-MS parameters), pattern shifts and response fluctuations are registered and evaluated to define a compensation strategy to be adopted in long-time-range studies. Results EVO oils volatiles patterns show misalignment both in 1D and 2D retention times (1tR, 2tR) and in peak widths, caused by chromatographic setup changes. For correct alignment, a supervised approach is designed , based on a targeted template of known analytes with their locations on the original 2D pattern; the template is then applied on the misaligned pattern and matching parameters iteratively changed to maximize the % of correct matches (distance threshold, SNR threshold, Mass Spectrum similarity to NIST libraries). By adopting optimized parameters, the automated procedure for feature template creation is tested for effectiveness. For both templates, i.e. targeted and featured untargeted one, the percentage of matching between misaligned patterns achieve the 95%. About response normalization, different approaches are evaluated (i.e. Internal Standardization ISTD, normalization on total image response or on partial image response): total image response normalization appears to be, on average, more effective, even if ISTDs normalization better compensates for response fluctuations on highly volatiles. Conclusions This study shows that, thanks to a careful optimization of “smart templates” parameters, it is possible to overcome 2D pattern misalignments and response fluctuations due to chromatographic parameters changes and MS system performances. A full metadata transfer together with univocal numbering/ID of targeted and untargeted analytes are obtained for both fully supervised fingerprinting and automated procedures (i.e. feature templates). Normalization on total image response is a good compromise to response normalization and enables reliable and consistent cross-sample analysis even with pattern misalignments due to chromatographic parameters changes.