No abstract

Lactose-free milk and milk substitutes based on soy, oat, or rice are widely marketed to lactose-intolerant consumers. In this study, 400 MHz 1 H nuclear magnetic resonance (NMR) spectroscopy was used in the context of food surveillance to validate the "lactose-free" claims labeled on these beverages. Using soft independent modeling of class analogy (SIMCA) analysis, a qualitative classification according to the type of beverage (lactose-containing milk, lactose-free milk, oat, soy, and rice milk substitutes) was possible. Furthermore, quantitative data regarding nutrition labeling parameters were predicted from the same spectra using partial least squares (PLS) regression. The models obtained for carbohydrate, sugars, protein, fat, saturates, and energy (R 2 =0.89-0.97) were suitable for a screening analysis. Using nicotinamide as an internal standard, quantitative determination of lactose with a detection limit of 0.03 g.L -1 was also possible using direct integration of the signals (linear range, 0.05-50.0 g.L −1 , R>0.999). The relative standard deviations for the lactose-free milks were below 10%. NMR spectroscopy was judged to be suitable for the rapid routine analysis of milk and milk substitutes.

Vegetable oils and fats may be used as cheap substitutes for milk fat to manufacture imitation cheese or imitation ice cream. In this study, 400 MHz nuclear magnetic resonance (NMR) spectroscopy of the fat fraction of the products was used in the context of food surveillance to validate the labeling of milk-based products. For sample preparation, the fat was extracted using an automated Weibull-Stoldt methodology. Using principal component analysis (PCA), imitation products can be easily detected. In both cheese and ice cream, a differentiation according to the type of raw material (milk fat and vegetable fat) was possible. The loadings plot shows that imitation products were distinguishable by differences in their fatty acid ratios. Furthermore, a differentiation of several types of cheese (Edamer, Gouda, Emmentaler, and Feta) was possible. Quantitative data regarding the composition of the investigated products can also be predicted from the same spectra using partial least squares (PLS) regression. The models obtained for 13 compounds in cheese (R 2 0.75–0.95) and 17 compounds in ice cream (R 2 0.83–0.99) (e.g., fatty acids and esters) were suitable for a screening analysis. NMR spectroscopy was judged as suitable for the routine analysis of dairy products based on milk or on vegetable fat substitutes.

In the original article two errors have been introduced into Chinese abstract.It concerns the following line copied here