Chemometric Classification of Apulian and Slovenian Wines Using 1 H NMR and ICP-OES Together with HPICE Data MARIA ANTONIETTA BRESCIA, † IZTOK J. KOS ˇ IR, ‡ VINCENZO CALDAROLA, † JURKICA KIDRIC ˇ ,* ,‡ AND ANTONIO SACCO § Dipartimento di Chimica, Campus Universitario, Universita ` di Bari, Via Orabona 4, 70126 Bari, Italy, National Institute of Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia, and Istituto di Produzioni e Preparazioni Alimentari, Facolta ` di Agraria, Universita ` di Foggia, Via Napoli 25, 71100, Foggia, Italy High-performance ion chromatography exclusion, inductively coupled plasma emission spectroscopy, and nuclear magnetic resonance (NMR) measurements were carried out in combination with chemometrics on 33 wine samples coming from three Slovenian wine-growing regions and from Apulia (southern Italy). The chemometric classification of wines according to their geographical origin was obtained with a nearly 100% degree of achievement. The discriminating potential of the 1 H NMR and of the other analytical determinations has been estimated separately. The best prediction of wines has been obtained with NMR data. KEYWORDS: Wines; NMR; ICP-OES; HPICE; chemometrics; geographical origin INTRODUCTION Wine is an important agricultural product which is, in many cases, a target of economic fraud, by, for example, the addition of exogenous sugars before or during the fermentation to increase the natural ethanol content in wine, the misleading of cheap wine, or the mixing of excellent and expensive wines with low-quality wines, often originating from other geographi- cal regions or even countries. Therefore, the determination of authenticity is a problem of increasing importance in the wine industry. Wine is a complex mixture consisting of several hundreds of compounds present at different concentrations. To carry out authenticity studies including determination of the chaptalization, geographical origin, and year of production, many compounds belonging to different chemical classes were tested using several different analytical techniques. For example, inorganic com- pounds were determined by flame emission spectrophotometry and atomic absorption spectrophotometry, and the determination of volatile compounds was done by gas chromatography-mass spectrometry (1). The analysis of trace elements was performed by inductively coupled plasma optical emission spectroscopy (2) and by electrothermal atomic absorption spectrometry (3). HPLC was used for the determination of amino acid composi- tion, organic acids, sugars, and polyphenols (4-6). The use of a combination of 1 H and 13 C NMR spectroscopy and pattern recognition techniques was also reported (7). Good results concerning the characterization of wine authenticity and the classification of wines according to the geographical origin were also obtained by means of stable isotope ratios determined by site-specific natural isotope fractionation NMR (SNIF-NMR) and isotopic ratio mass spectrometry (IRMS) (4, 9, 10, 11). The application of different analytical methods for the deter- mination of many components in wine results in large amounts of data. The use and the evaluation of so many parameters are common problems in food chemistry. To deal with this problem, chemometrics have been applied to different chemical or physical variables in the past decade. Principal component analysis (PCA), soft independent model- ing of class analogy (SIMCA), and Kohonen artificial neural networks (KANN) were used on typical variables determined in wines by classical wine analysis. KANN was employed for the classification of wines or wine vinegars using the concentra- tion of trace elements (1, 2, 11, 12). PCA and discriminant analyses (DA) were applied on proton and 13 C chemical shifts and signal intensities (7). PCA, linear discriminant analysis (LDA), KANN, and cluster analysis were used on SNIF-NMR and IRMS results (8-10). In the present work, we compare the use of different analytical techniques for the determination of different compounds in wine in combination with chemometric methods for the classification of wines from Slovenia and from Apulia - a region in the south of Italy. The analytical techniques used to obtain the input parameters for chemometric analysis were high-performance ion chromatography exclusion (HPICE) for the determination of some organic acids, inductively coupled plasma emission spectroscopy (ICP-OES) for the determination of trace elements, and 1 H nuclear magnetic resonance (NMR) for the semiquan- titative determination of some amino acids, organic acids, and alcohols. The data obtained by classical analytical methods and * Corresponding author: Tel., +386 01 4760272; fax, +386 01 4259244; e-mail, jurka.kidric@ki.si. † Universita ` di Bari. ‡ National Institute of Chemistry. § Universita ` di Foggia. J. Agric. Food Chem. 2003, 51, 21-26 21 10.1021/jf0206015 CCC: $25.00 © 2003 American Chemical Society Published on Web 11/26/2002