Optimization and classification of natural products by chemometric method. Analysis of gas chromatographic data of essential oils of Indonesia Myristica fragans Houtt.

Essential oils (EO) of nutmeg (Myristica fragan Houtt) have a different chemical composition according to their origin. So it is important that laboratories of quality control have a fast and effective method of analysis to prevent frauds and adulteration. The analytical method generally employed [1], is time consuming and two pairs of compounds were not resolved (sabinene/β-pinene, p-cimene/limonene). In order to optimize the GC separation, while reducing the analysis time, we have identified the influential analytical factors among 11 by using a screening of factors: 5 factors are influent. After, we studied the interactions between these 5 factors by using a fractional factorial experiment matrix 2^{5 – 1} in 16 experiments [2] and by considering the interactions effect of first order b_ij. This optimization allowed to reduce analysis time (20 min instead of 55 min) and to improve the separation between the unresolved compounds (Resolution > 1.5). This method was applied to differentiate EO from ripe nutmeg and those from maces: a chemometric treatment (PCA) of GC data was carried out. The repartition of samples in a 3D score plot (PC1; PC2; PC4 explained variance 81%) show two groups: the maces and the ripe nutmeg. The study of loading associated to these compounds showed that the EO of maces are different on the basis of their content of phenylpropenic compounds (eugenol, myristicin, safrol...) while the EO of ripe nutmegs is characterized by the content of terpenic compounds (α and γ-terpinene, sabinene, α and β-pinene...). However, an overlap of the different classes exists. So the supervised classification method SIMCA [3] was used. Four samples of each origin were used to build the disjoint PCA and three from the same origin to test the models: all the samples are well classified. This study allowed to reduce analysis time and to improve the separation between all the compounds, and to differentiate EO from their origin: this method can be used as quality control.

Keywords: Gas Chromatography (GC), Myristica fragans, fractional factorial design, Principal Component Analysis (PCA), Soft Independent Modeling of Class Analogy (SIMCA).

References:

[1] AFNOR 2007. Norme Huile essentielle de muscade, type Indonésie (Myristica fragrans Houtt.). NF ISO 3215. Dans Huiles essentielles Tome 1: Échantillonnage et méthodes d'analyse, AFNOR (Association Française de Normalisation), Paris

[2] Cela R, Claeys-Bruno M, Phan-Tan-Luu R. Screening strategies in comprehensive chemometrics, Brown TR, Walczak S. Oxford: Elsevier 2009; 251 – 300

[3] Dupuy N, Le Dréau Y, Ollivier D, Artaud J, Pinatel C, Kister J. Origin of French virgin olive oil registered designation of origins predicted by chemometric analysis of synchronous excitation-emission fluorescence spectra. J Agric Food Chem 2005; 53: 9361 – 9368