Towards harmonization of non-targeted 1H NMR spectroscopy-based wine authentication: Instrument comparison. (February 2022)

• Record Type: Journal Article
• Authors: Ehlers, Mona; Horn, Bettina; Raeke, Julia; Fauhl-Hassek, Carsten; Hermann, Armin; Brockmeyer, Jens; Riedl, Janet
• Is Part Of: Food control. Volume 132(2022)
• Publication Date: 2022-02
• Subjects: Food authenticity -- NMR-Based metabolomics -- Instrument comparison -- Grape variety classification -- Multivariate statistical analysis -- Jointly used database; swRSD spectrum-wide relative standard deviation; Food -- Quality -- Periodicals; Food -- Analysis -- Periodicals; Food handling -- Periodicals; Food industry and trade -- Quality control -- Periodicals; Aliments -- Industrie et commerce -- Qualité -- Contrôle -- Périodiques; Aliments -- Qualité -- Périodiques; Aliments -- Analyse -- Périodiques; Hygiène alimentaire -- Périodiques; Food -- Analysis; Food handling; Food -- Quality; Periodicals; Electronic journals; 664.07
• DOI: 10.1016/j.foodcont.2021.108508
• Languages: English
• ISSNs: 0956-7135

Abstract

Abstract: Non-targeted Nuclear Magnetic Resonance (NMR) spectroscopy is a promising approach for wine authentication. A prerequisite for its application in official control is the inter-laboratory comparability, which has rarely been the subject of research studies so far. For the first time, in this study an instrument comparison was carried out in order to investigate the extent and impact of instrumental differences on the classification of wine grape varieties. Following a harmonized protocol for sample preparation, spectra acquisition and processing, the same sample set, comprising of 101 red and 100 white wine samples, was analyzed with two NMR spectrometers of the same vendor in different laboratories. Observed spectral differences (intensity, baseline, chemical shift of pH-sensitive signals) were accounted for by applying an instrument-specific conversion factor or normalization to total intensity, respectively. Partial Least Squares – Discriminant Analysis (PLS-DA) and Principal Component Analysis – Linear Discriminant Analysis (PCA-LDA) classification models built with data from the individual instruments and with combined data from both instruments achieved comparable accuracies in the internal validation (79–87%). When classification models of the individual instruments were used to predict data of the respective other instrument, results were inconsistent, whereas the use of combined data from both instruments for training and test sets produced more robust and reliable results. This study represents an important step towards harmonization of non-targeted analytical approaches and may help in the development of jointly used databases for food authentication. Highlights: Instrument comparison of non-targeted 1 H NMR spectroscopy for wine authentication. Comprehensive study with 201 commercial wines covering natural product variation. Reduction of inter-laboratory variation by novel conversion factor or normalization. Both instruments achieved comparable results for grape variety classification. Model based on data of both instruments produced most consistent prediction results.