Mechanism-based markers for early assessment of lipid oxidation in mayonnaise by magnetic resonance spectroscopy

Lipid oxidation is a major concern in the food industry as it is one of the main contributors to the limited shelf life of lipid-based food products. This is a particular concern for food emulsions with high amounts of unsaturated fats, such as mayonnaise. In the food industry, accelerated shelf-life tests (ASLTs) are often applied to assess the oxidative stability of different formulations. These ASLTs are not used without drawbacks. First, it is well recognised that shelf-life testing under accelerating conditions may induce different mechanisms and therefore, different stability rankings than when performed under consumer-relevant conditions. Secondly, even under accelerated conditions, it can still take weeks of shelf-life before (late) off-flavour compounds can be detected. Thirdly, it is difficult to infer the involved mechanism from these late lipid oxidation products. To optimise formulation and processing of oxidation-stable food emulsions in a more rapid and rational manner there is a clear need for shelf-life tests that provide early (hours-days) markers that quantitatively predict generation of off-flavour and give an indication of the involved oxidative mechanism(s). The main goal of the thesis is to discover these mechanism-based early markers for lipid oxidation in food emulsions, possibly under consumer relevant conditions.An Nuclear Magnetic Resonance (NMR) method for the quantitative assessment of both hydroperoxides and aldehydes, respectively the primary and secondary oxidation products, was developed using band-selective selective 1H NMR pulse excitation. 1H NMR signals of hydroperoxides were assigned in a fatty acid and isomer specific way. The method allowed quantification of hydroperoxides and aldehydes with high throughput, dynamic range and precision. Explorative multivariate data modelling of the quantitative 1H NMR profiles revealed that shelf-life temperature had a significant impact on lipid oxidation mechanisms. The method was expanded to include the epoxides, using 2D 1H-13C HSQC. This addition allowed for the inclusion of also the major epoxide contribution in the assessment of environmental impact on lipid oxidation pathways. Different classes of epoxides were distinguished, based on stereospecificity and additional oxidation moieties. Limited potential for epoxides as a sole early marker was found, however.&nbsp; Combined assessment of&nbsp; hydroperoxides and epoxides will allow for capturing of main radical mechanisms.Radicals were monitored in early lipid oxidation using Electron Spin Resonance (ESR), combined with spin-trapping. By following spin adducts over time, oxidative stability of emulsion could be derived. Importantly, already at mild (=<50 ºC) shelf-life conditions, the PBN-OOL adducts rapidly degraded to MNP, benzaldehyde and alkoxy-radicals. In contrast to what is often believed, the observed ESR-signal predominantly originated from the adducts of MNP and alkyl radicals. The trapping of LOO∙ radicals and subsequent generation of LO∙ radicals impacted downstream reaction pathways and hereby precluded mechanistic studies in the presence of PBN.An extensive set of quantitative hydroperoxide and aldehyde data from a wide range of emulsion formulations under accelerated shelf-life tests (50 ºC) was used to train a prediction model for late oxidation kinetics. The initial phase of hydroperoxide generation could be described by a sigmoidal Foubert-adapted model. The point at which this function reached the proposed CCLOOH of 38 mmol/kg was used as an accurate predictor for the onset of aldehyde generation in mayonnaise formulations. LOOH measurements within a few days of shelf-life are sufficient to accurately predict the aldehyde onset which takes place at a time scale of weeks. Furthermore, the parameters that describe the Foubert function, i.e., the induction delay (f0) and the growth rate (K), could be used to discriminate primary and secondary anti-oxidative mechanisms.Finally, a 31P NMR based method was developed to quantify loading of phosvitin with Fe(III) and its reductive release. Both features could be quantified in model phosvitin solutions by exploiting the paramagnetic broadening of 31P NMR signal of phosphoserine residues by Fe(III). This method was then successfully applied to quantify the phosvitin-Fe(III) loading in mayonnaise water phase by liquid NMR, whereas 31P NMR MAS could only provide a qualitative measure. The 31P NMR method showed that lowering the Fe(III) loading of phosvitin in mayonnaise by EDTA-chelation directly correlated with lipid oxidation stability. The strengths and limitations of the presented methods are reflected upon for both spectroscopic and modelling methods. The effect of mechanistical markers on the interpretation of accelerated shelf-life tests and the pivotal role of hydroperoxides in mayonnaise are then discussed, followed by the wider applicability of the provided methods, including other oil types and matrices. Lastly, future perspectives to further elucidate the physico-chemical interplay in emulsions and prediction of sensorial properties are outlined.