Screening method for determination of C18:1 trans fatty acids positional isomers in chocolate by 1H NMR and chemometrics

• Published in: Food science & technology 2020-09, Vol.131, p.109689, Article 109689
• Main Authors: Amorim, Tatiane Lima, Duarte, Lucas Mattos, Granato, Álisson Silva, de Oliveira, Marcone Augusto Leal, Amarante, Giovanni Wilson, de la Fuente, Miguel Angel, Gómez-Cortés, Pilar
• Format: Article
• Language: English
• Subjects: Gas chromatography; Industrial trans fat; Multivariate calibration; Nuclear magnetic resonance; Partial least square regression
• ISSN: 0023-6438; 1096-1127
• DOI: 10.1016/j.lwt.2020.109689

The regulations established worldwide, restricting trans fatty acids (TFA) from industrial origin in foods, encourage the development of rapid methods to verify the quality of food products. A simple and rapid procedure based on 1H NMR combined with multivariate calibration was developed to quantify C18:1 TFA isomers in chocolate fat. Thirty chocolate samples (milk, dark and white chocolates) were analyzed by 1H NMR, and also by a reference gas chromatographic method. Then, partial least squares regression was used for model development. The most important 1H NMR signals for the modeling comprised the 5.25–5.45 ppm region, corresponding to CH (double bonds and glycerol backbone), and the 1.94–1.99 ppm region, characteristic of CH2 allylic to trans double bonds. Good predictive qualities for all TFA isomers were obtained. Coefficient of correlation for prediction was 0.967 for C18:1 trans 6–8; 0.981 for C18:1 trans 9; 0.987 for C18:1 trans 10; 0.941 for C18:1 trans 11; 0.983 for C18:1 trans 12 and 0.977 for total TFA. This study demonstrates the potential of 1H NMR to be applied in food industries in order to perform TFA control analysis as well as to investigate food fraud. •TFA positional isomers and total TFA contents were predicted in chocolates.•1H NMR and partial least square regression models were used for modeling.•TFA estimations by 1H NMR + PLSR were well correlated to GC results.•This approach can be useful for TFA monitoring by industries and for fraud detection.