Files
Abstract
Lipid autoxidation is one of the central economic concerns to the edible oil and food & beverage industry. The chemistry of this degradation reaction has been understood for many decades, but successful means to predict or optimize oxidative stability have proved elusive. It is speculated here that an ongoing hindrance to these efforts has been the lack of consistency of assay selection, study design, and numerical interpretation technique among scientists attempting to quantify oxidative stability within fats and oils. This study monitored 50 samples of current commercial-use edible fats and oils for the accumulation of lipid autoxidation products according to four distinct assays throughout two months of accelerated storage. This oxidation data was examined for the derivation of a single novel comprehensive quantification of a samples exhibition of oxidative stability. This comprehensive term was then used as the basis for the development of predictive models of oxidative stability according to numerous composition factors. Sample unsaturation (as concentrations of monounsaturated fatty acids, diunsaturated fatty acids, and triunsaturated fatty acids) demonstrated a strong correlation (R2 = 91.5%) with the stability term, and indicated the combined presence of multiple double bonds on individual fatty acids to be associated with impaired oxidative stability. A systematic sequential approach to model-building was then employed to negotiate the challenges of inherent redundancies within the composition variables of edible fats and oils. Independent of sample unsaturation, triacylglycerols containing one, two, and seven double bonds were positively associated with stability, and the concentrations of triacylglycerols containing three, four, five, and six double bonds were negatively associated with stability. trans-Fatty acids, sample purity, and -tocotrienol were also associated with improved stability. Unsaturated fatty acids greater than 18 carbons in length and -tocotrienol were both associated with impaired stability. Final models including considerations of these factors were highly predictive of oxidative stability (R2adj = 97.1% for oil blends and R2adj = 96.2% for pure samples).