|JABLONSKI, JOSEPH - Us Food & Drug Administration (FDA)|
|MOORE, JEFFREY - Us Pharmacipeia (USP)|
|Harnly, James - Jim|
Submitted to: Journal of Agricultural and Food Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/8/2014
Publication Date: 5/8/2014
Citation: Jablonski, J., Moore, J., Harnly, J.M. 2014. A non-targeted UHPLC-UV methid with classical and multi-variate data analysis to detect adulteration of skim milk powder with foreign proteins. Journal of Agricultural and Food Chemistry. 62:5198-5206.
Interpretive Summary: A method was evaluated for the detection of skim milk powder (SMP) adulterated with foreign proteins using ultra high performance liquid chromatography. The foreign proteins were soy (SPI), pea (PPI), brown rice (BRP), and hydrolyzed wheat protein (HWPI) isolates added to the SMP at levels of 0.5%, 1%, 3%, and 10%. The data for the authentic and adulterated SMP samples were examined using a pattern recognition software and by a simple statistical test applied to the authentic and adulterated samples at each retention time. These tests showed that adulteration with SPI and PPI were detectable at the 0.5 to 1.0% level while BRP and HWPI were detectable at 3.0% and 10% levels.
Technical Abstract: Ultra-High performance liquid chromatography (UHPLC) with single wavelength (215 nm) detection was used to obtain chromatographic profiles of authentic skim milk powder (SMP) and synthetic mixtures of SMP with variable amounts of soy (SPI), pea (PPI), brown rice (BRP), and hydrolyzed wheat protein (HWPI) isolates. Preparations containing 10% and 3% SPI, PPI and BRP provided clearly visible unexpected peaks in the SMP chromatograms. Profiles of authentic and adulterated samples were compared at the 0.5, 1.0, 3.0 and 10% level using univariate and multivariate statistics. The univariate method determined the 99% confidence limit for each chromatographic point based on the authentic SMP samples. Adulteration was detected based on the number data points exceeding the 99% confidence limit in the chromatograms. The multivariate approach used the Q statistic from SIMCA fit to authentic samples to determine adulteration. In general, both approaches gave similar results. SPI and PPI were detectable at the 0.5 to 1.0% level while BRP and HWP were detectable at 3.0% and 10% levels, respectively.