1H NMR Adulteration Study of Hempseed Oil of Different Geographical Origin Using Interval and Merged-interval Regression Procedures

Adulteration of high-priced, health beneficial hempseed oil of different geographical origin was studied by combination of 1H NMR spectroscopy and multivariate statistical analysis. The fatty acid composition of hempseed oil corresponds to the ideal ratio of essential fatty acids (EFAs) required by human body, which is roughly 3:1 of omega-6 to omega- 3. 15 hempseed oil samples were adulterated with 6 samples of sunflower oil, 4 samples of rapeseed oil and 4 samples of sesame oil. Each particular edible oil sample varies in producer and country of origin. Including binary mixtures there were altogether 105 oil samples. The aim of this study was to determine the content of each studied hempseed oil sample using established methodology regarding iodine value [1] and fatty acid composition [2], and to select optimal variables and establish optimal regression model for prediction of adulteration with already mentioned adulterant oils. Obtained results show that although hempseed oil samples vary considerably in iodine value (154.0 − 165.5) and omega-3 fatty acids (15.7 ‒ 20.0 %), interval ridge regression [3] and first-break forward interval partial-least square regression (FB- FiPLS) [4] obtain significantly lower root-mean error of prediction than PLS applied to the whole considered NMR spectral region (6 ‒ 0 ppm) in most studied cases. To the best of our knowledge merged-interval regression procedures are for the first time applied to NMR data. Obtained prediction accuracy for volume fraction of each adulterant oil in binary mixtures (1.4 ‒ 3.0 %, 0.991 < R2 < 0.998) are promising enough to conclude that 1H NMR combined with the inspected chemometric procedures can be used to effectively quantify adulteration in hempseed oils, even when many samples of different geographical origin are considered for both hempseed oil and adulterant oils. [1] Y. Miyake, K. Yokomizo, N. Matsuzaki, J. Am. Oil Chem. Soc. 1998, 75, 15. [2] R. Sacchi, F. Addeo, L. Paolillo, Magn. Reson. Chem. 1997, 35, S133. [3] O. Jović, N. Smrečki, Z. Popović, Talanta 2016, 150, 37. [4] O. Jović, Food Anal. Methods 2016, 9, 281.