AOAC 2018 Methods

where c = concentration of analyte in the test sample (expressed as free 3-MCPD, 2-MCPD and glycidol equivalent, respectively, mg/kg sample); A a = area of the peak corresponding to the analyte; IS = absolute amount (in µg) of internal standard added to the test sample; A IS = area of the peak corresponding to the Internal Standard; a = slope of the calibration curve; W = weight of the sample (in g); b = y-intercept of the calibration curve

J. Data Handling

Report the result of 3-MCPD, 2-MCPD, 3-MCPD esters, 2-MCPD esters and glycidyl esters in mg/kg to 4 decimal places.

K. Performance Characteristics

(a) Samples for single lab validation . – Two powdered and two liquid infant formula/adult nutritionals were selected to determine the performance parameters of the analytical method. Samples were selected from an internal survey on the basis of their MCPD/GE levels as close as possible to the LOQ. In absence of certified reference material, 5 samples from SPIFAN II kit were selected for duplicate analysis because of their incurred levels of MCPD/GE. Samples considered are described in Table 2018.03E . (b) Fat extraction . – Scope of the method is Infant formula and adult nutrition. This matrix category is known to be wide in terms of formulation, with a fat content highly variable. For powder formula, structure is highly dependent on ingredients and manufacturing process, leading to microstructure of the matrix that is unique. Fat is encapsulated, and its extraction is challenging. Fat extraction method considered in this method is a modified and simplified version of fat extraction method published earlier by Leigh et al. [1]. This approach has been tested on a wide scope of Infant and adult nutrition formula (n=29 different recipes), comprising of powder and liquid samples. Results were compared to the reference method used for fat content labelling (Rose Gottlieb). As shown in Table 2018.03F , fat extracted for all samples does not differ by more than 10% from the reference method. Statistical comparison of results generated by both methods, shows that the median of differences is -0.67 and correlation between the two sets of data has a slope of 0.99 (confidence interval [0.95;1.04] includes 1), and an intercept of -0.53 (confidence interval [-1.7;0.61] includes 0). (c) LOQ. – LOQ was set at the lowest level of calibration, as the first calibration level fulfill the following criteria: reproducibility with a precision of 20% and an accuracy of 80%–120% (as shown in Linearity/ Calibration fit Table 2018.03G ). LOQ are summarized in Table 2018.03H . Those low LOQ can be confirmed also by the results generated on samples with low native values. As shown in Table 2018.03I under precision studies, a liquid sample (adult Nut. 3 liquid) was analyzed for 2- MCPD ester at 1.2 μg/kg, with an RSD r and RSD iR of 19.3 and 23.5%, respectively. (d) Linearity/ Calibration fit. – Calibration curves were plotted and linearity demonstrated by R2 > 0.99. Back calculated concentration were within 80-120% of their nominal value. At the lowest concentration level (0.002 µg for esters form, 0.005 µg for free form), back calculation accuracy are showed in Table 2018.03 G as well as reproducibility (<10%) calculated over six different occasions.