Fast analysis of some phenolics by hplc-dad and chemometric resolution techniques

In recent years, the prevention of cardiovascular diseases and cancer has been intensively focused on the beneficial effects of olive oil and, in particular, of its phenolic pool. These species are powerful inhibitors of the formation of atherosclerotic plaques and reactive oxygen species, cause of "fat-related" cancers. It seems clear that the phenolic composition represents a specific index of nutritional quality of the product and how important it is to develop rapid, economic and simple analytical methods for its determination in olive oils. To this end, since chemometrics allows the mathematical a posteriori resolution of overlapping peaks, it is possibile to operate with experimental conditions even leading to a not perfect chromatographic separation, with a corresponding saving of money and time and reduction in the amount of solvents used. Accordingly, in this communication, the possibility of operating a rapid HPLC-DAD analysis of a wide range of polyphenols, using a fast gradient involving two solvents only (MeOH/H2O) and chemometric multivariate and multiway techniques, will be discussed. In particular, two approaches were tested for the resolution of peak clusters: the use of Multivariate Curve Resolution (MCR [1]) after unfolding of the experimental data cube and the application of PARAFAC2 [2] directly on the three way array. The results obtained by both methods have been validated by comparison with a complete chromatographic resolution of the co-eluted peaks. In particular, it was shown that both chemometric approaches led to good results but MCR performed better on the investigated data set. A possible explanation of this outcome could be found in the fact that, as the spectral profiles of the analytes are very similar, further constraints are needed to separate the contribution of the individual components, for instance unimodality of the peaks, and these constraints can’t be applied in PARAFAC2 due to algorithmic reasons.