Characterization Of Milk And Detection Of Milk Fat Adulteration By Fluorescence Spectroscopy Coupled With Chemometrics

A number of methods have been developed to date to test the authenticity and detectadulteration of commercial food products. However, some of these techniques are not simple touse and are labor intensive, and also involve very high running costs, thereby hamperingapplication of such techniques in monitoring programs at large scale. Contrarily, spectroscopicmethods yield information on all the components of a mixture in one spectrum, and usuallywithout the need for sample pretreatment. In the present study, the potential application offluorescence spectroscopy combined with chemometric techniques to characterize commercialmilk and detect the adulteration of milk fat with vegetable oil was assessed. The study wasconducted in three stages involving characterization, discrimination and detection.International and domestic milk brands comprised of pasteurized and ultra hightemperature (UHT) milks of different composition (whole, reduced fat, skimmed, low lactose,and high protein) were obtained from the local supermarkets. The emission spectra of Maillardreaction products (MRP) and riboflavin (RF), and tryptophan were recorded using front facefluorescence spectroscopy (FFFS). Principal component analysis (PCA) was applied to analyzethe normalized spectra. PCA of the spectral information from MRP/RF discriminated the milksamples from various countries, and PCA of spectral information from tryptophan discriminatedthe samples according to their compositions. The fluorescence spectral data were compared withliquid chromatography-mass spectrometry (LC-MS) results for the extent of protein glycation inthe milk samples, and a positive association (R~2=0.840) was found between the degree ofglycation of α-lactalbumin and the MRP/RF spectral data.Fluorescence spectroscopy was also used to discriminate between the fat of animal(butterfat) and plant origin (vegetable oil) and determine their fatty acid profiles. The followingvegetable oils were used; sunflower, maize, rice-bran, peanut, soybean, virgin olive, tea/camellia,sesame, canola and blended oil. The3-dimensional (3D) and vitamin E fluorescence spectrawere recorded from the samples, and the fatty acid profile was determined by gaschromatography (GC). Multivariate (PLS2) and three-way (3-way PLS) partial least squaresregression analysis were applied to the vitamin E and3D data respectively. The butterfat wasclearly discriminated from the vegetable oils, and the various oils were also discriminated fromone another according to first three principal components which accounted for>96%of the total variance. The butterfat and vegetable oils were separated based on total saturated andunsaturated fatty acids (SFA&UFA) respectively, while tocopherols and tocotrienols accountedfor greater variability among the various oils. A very good prediction model was described withhigh coefficients of determination (R~2) within acceptable error limits when using vitamin Espectra than using the3D spectra. The myristic, palmitic and stearic acids displayed R~2of0.992,0.745&0.945respectively, while the UFA displayed R~2<0.339.The fluorescence spectroscopy was also applied to detect adulteration of butterfat andcommercial milk with vegetable oil, and identify the source of their fat. Pure butterfat wasadulterated with vegetable oils at various concentrations (0,5,10,15,20,30,&40%), and thenon-fat and reduced fat milks were adulterated with vegetable oil to simulate whole fat milk oftwo kinds. Pure commercial milk of different brands&composition was included in the analysisas the control. The2D&3D FFFS and GC were used to obtain fluorescence spectra and fattyacid profiles respectively. PCA&PLS2were applied to analyze the data. The butterfat sampleswere discriminated according to the first two principal components which accounted for98%ofthe total data variability, and these correlated to the total SFA and C18:2, C18:1&C18:3acidswhich were present in high concentrations in pure butterfat, sunflower and canola adulteratedbutterfat samples respectively. Adulteration of butterfat with vegetable oil at5%was detected.There was clear identification of the three sample groups of milk. The samples werediscriminated according to the first three principal components, and these correlated to threemajor fluorophores present in the samples; riboflavin, tryptophan and vitamin E. Among thefatty acids, C18:2&C16:0acids accounted for greater portion of the variability among thesamples. The SFA displayed higher predictability than the UFA,(R~2=0.729–0.907and0.202–0.650respectively) with C18:1&C18:3being the poorly predicted fatty acids. This studydemonstrated the ability and sensitivity of FFFS to rapidly characterize commercial milk ofvarious compositions and processing conditions, and differentiate between butterfat andvegetable oils, and predict their SFA composition. Adulteration of milk fat with vegetable oilwas also detected.