Karunathilaka Sanjeewa R, Farris Samantha, Mossoba Magdi M
U.S. Food and Drug Administration, Center for Food Safety and Applied Nutrition, Office of Regulatory Science.
J Oleo Sci. 2018 Dec 1;67(12):1501-1510. doi: 10.5650/jos.ess18098. Epub 2018 Nov 15.
The ruling that partially hydrogenated oils (PHO) are no longer "generally recognized as safe (GRAS)," has accelerated the replacement of PHO ingredients with fat alternatives having increasingly lower or no trans fat content. In the present study, we developed a Fourier-transform infrared (FTIR) spectroscopic procedure in conjunction with multivariate partial least squares regression (PLSR) and found it suitable for the accurate prediction of low (0.5%) total trans fat content, as percentage of total fat, measured as fatty acid methyl esters (FAME), in the lipids extracted from 24 representative fast foods. This multivariate data analysis approach is relevant because the precision of the current univariate FTIR official method (AOCS Cd 14-09) is reportedly poor below 2% of total fat, while PLSR has allowed us to accurately predict the concentration of low trans fat in fast foods. The performance of a portable FTIR device was also evaluated and compared to that of a benchtop FTIR spectrometer. For both infrared data sets, PLSR-predicted concentrations of total trans FAME, ranging from approximately 0.47% to 11.40% of total FAME, were in good agreement with those determined by a primary reference gas chromatography (GC) method (R>0.99); high prediction accuracy was also evidenced by low root mean square error of cross-validation (RMSECV) values. The lowest RMSECV error of 0.12% was obtained with the portable device. The lowest total trans FAME concentration, determined by GC to be 0.42%, was accurately predicted by the portable FTIR/PLSR procedure as 0.47% of total FAME.
部分氢化油（PHO）不再被“普遍认为是安全的（GRAS）”这一裁定，加速了用反式脂肪含量越来越低或不含反式脂肪的脂肪替代品来替代PHO成分的进程。在本研究中，我们开发了一种结合多元偏最小二乘回归（PLSR）的傅里叶变换红外（FTIR）光谱法，发现它适用于准确预测从24种代表性快餐中提取的脂质中低含量（0.5%）的总反式脂肪含量（以总脂肪酸甲酯（FAME）形式测量，占总脂肪的百分比）。这种多元数据分析方法具有重要意义，因为据报道，当前的单变量FTIR官方方法（AOCS Cd 14 - 09）在总脂肪含量低于2%时精度较差，而PLSR使我们能够准确预测快餐中低反式脂肪的浓度。我们还评估了便携式FTIR设备的性能，并将其与台式FTIR光谱仪的性能进行了比较。对于这两个红外数据集，PLSR预测的总反式FAME浓度范围约为总FAME的0.47%至11.40%，与通过主要参考气相色谱（GC）方法测定的浓度高度一致（R>0.99）；交叉验证的低均方根误差（RMSECV）值也证明了高预测准确性。便携式设备获得的最低RMSECV误差为0.12%。通过GC测定的最低总反式FAME浓度为0.42%，便携式FTIR/PLSR程序准确地将其预测为总FAME的0.47%。
