Pattern recognition analysis of chromatographic fingerprints of Crocus sativus L. secondary metabolites towards source identification and quality control.

Chromatographic fingerprinting is an effective methodology for authentication and quality control of herbal products. In the presented study, a chemometric strategy based on multivariate curve resolution-alternating least squares (MCR-ALS) and multivariate pattern recognition methods was used to establish a gas chromatography-mass spectrometry (GC-MS) fingerprint of saffron. For this purpose, the volatile metabolites of 17 Iranian saffron samples, collected from different geographical regions, were determined using the combined method of ultrasound-assisted solvent extraction (UASE) and dispersive liquid-liquid microextraction (DLLME), coupled with GC-MS. The resolved elution profiles and the related mass spectra obtained by an extended MCR-ALS algorithm were then used to estimate the relative concentrations and to identify the saffron volatile metabolites, respectively. Consequently, 77 compounds with high reversed match factors (RMFs > 850) were successfully determined. The relative concentrations of these compounds were used to generate a new data set which was analyzed by multivariate data analysis methods including principal component analysis (PCA) and k-means. Accordingly, the saffron samples were categorized into five classes using these techniques. The results revealed that 11 compounds, as biomarkers of saffron, contributed to the class discrimination and characterization. Eleven biomarkers including nine secondary metabolites of saffron (safranal, α- and β-isophorone, phenylethyl alcohol, ketoisophorone, 2,2,6-trimethyl-1,4-cyclohexanedione, 2,6,6-trimethyl-4-oxo-2-cyclohexen-1-carbaldehyde, 2,4,4-trimethyl-3-carboxaldehyde-5-hydroxy-2,5-cyclohexadien-1-one, and 2,6,6-trimethyl-4-hydroxy-1-cyclohexene-1-carboxaldehyde (HTCC)), a primary metabolite (linoleic acid), and a long chain fatty alcohol (nanocosanol) were distinguished as the saffron fingerprint. Finally, the individual contribution of each biomarker to the classes was determined by the counter propagation artificial neural network (CPANN) method.