Show me its volatile profile and I will tell you its quality: rapid fingerprinting of food volatilome at the boundary between food industry and sensory perception

Summary Background and aim: Food quality is a multidimensional concept which includes both objective and subjective factors. Among these factors, the sensory aspect is gaining more and more attention among consumers and food industry. Monitoring this aspect of quality along the food chain, and in particular the one linked to product flavour, have become essential to ensure competitiveness. In this context, food industries are searching for rapid and flexible instrumental tools to support sensory evaluations of products. Monitoring volatile organic compounds (VOCs) from farm to fork through direct injection mass spectrometry techniques seems a promising possibility. This thesis focuses on the aspects of food quality related to the sensory part of aroma by presenting different applications of proton transfer reaction mass spectrometry (PTR-MS). The technique was applied to monitor VOCs changes during products shelf life, for control and investigation of raw materials’ quality and to investigate drivers of aroma release and perception during consumption. Material and methods: PTR-ToF-MS was used in combination with a multipurpose autosampler for obtaining, through a rapid and non-invasive headspace analysis, VOCs fingerprint of different food matrixes like anhydrous milk fat (AMF), ultra-high temperature lactose-free milk (UHT LF milk) and raw hazelnuts. H3O+ was mainly used as precursor ion for VOCs ionization but by using a selective reagent ionization device (SRI) the possibility to use other ionization ions (NO+ e O2 + ) was explored to obtain additional information for compound identifications and for increasing samples discrimination. As well, the introduction of an ion funnel at the end of drift tube, allowed to increase technique sensitivity by improving ions transmission from the drift tube to the mass analyser region. A similar improvement in sensitivity, was obtained when using a PTR-QiToF-MS for in vivo nose-space analysis. The instrument was combined with a heated device for sampling participant’s breath from nostrils which has a double advantage. On one hand it improves panellists’ comfortability and, on the other hand, it improves transmission of all VOCs from the device to the drift tube by reducing VOCs adsorption phenomena on tubing surfaces. To correlate volatilome to perceived quality, sensory data were also collected. For raw materials quality control industrial sensory evaluation tests were performed, mostly through “A” – “not A” test where samples are compared to an industrial golden standard. For in vivo experiments, a Time Intensity method was used to follow aroma perception of panellists. Different type of recruiting criteria where included for participants. When the focus was on characterizing product reformulation in mayonnaise and the effect of combining it with different carrier foods (bread/potato) on aroma perception and release, only Dutch young females were recruited to limit inter-individual variability. For the same reason an eating protocol (fixed chewing time and swallowing) was established. When the focus was on characterizing inter-individual variability, and how gender, ethnicity and physiological parameters affect aroma perception and release, a panel constituted of Appendix 225 Appendix 225 Asian-Chinese and Caucasian-Europeans (both males and females) was recruited. No fixed chewing procedure was imposed when consuming a mint chewing gum without any aroma encapsulation and sugar coating. Results: The results of this thesis confirm that PTR-ToF-MS fingerprinting is a valid sensitive approach to collect information about VOCs evolution during shelf life at different conditions and to monitor degradation processes which can compromise food quality such as lipid oxidation. In Chapter 2, the proposed methodology allowed to identify the effect of different production variables on AMF volatilome. Packaging type increased AMF volatilome differences during shelf life at 4°C, while storage at 50°C decreased them due to thermal oxidation phenomena that led to more elevated intensities of aldehydes and (methyl)ketones. Production batches differences also decreased during storage at 50°C. These VOCs differences were highlighted as well in UHT LF milk in Chapter 4. Most mass peaks increased during shelf life and differences in the batches were found in early shelf life stages. In both the experiments (Chapter 2 and 4), the rapidity of the analysis allowed to include and assess simultaneously more variables. Results from Chapter 3 and 5 demonstrate that VOCs fingerprints obtained with PTR/SRI-ToF-MS, coupled with a multipurpose autosampler to enhance analysis standardization and computing power, can be used for supporting sensory quality control in agroindustry. For both tested raw materials (AMF and raw hazelnuts), the good quality samples had lower concentrations for most of the detected mass peaks in VOCs fingerprints. This is in agreement with sensory results obtained by the industrial partner where the “gold” standard for a raw material is to have a flavour as neutral as possible, without any off-notes. Both unsupervised and supervised data mining approaches gave positive results in discriminating sensory classes. In Chapter 3, PLS-DA predictive models on AMF showed a correct classification rates of 97% in discriminating good quality samples from non-conform samples (data form NO+ ionization mode) while unsupervised clustering on raw hazelnuts (Chapter 5) had a correct classification rate of 95% in separating conform and non-conform samples (data from H3O+ mode). Therefore, in an industrial quality control program, it will be possible to test the quality of a higher samples number through their volatile fingerprint obtained with fast and non-invasive PTR-MS approach. Industrial sensory evaluation will be performed only on samples discarded by the instrumental measurement to confirm the defected quality. Moreover, in Chapter 5 the approach ability to discriminate between raw hazelnuts lots with different levels of gustatory and visual defects was tested. The technique sensitivity was enough to detect significant VOCs variations between samples with 10 and 20% of defected hazelnuts. Testing for defects in large lots of raw materials in few seconds represent another improvement for industrial quality control programs. Chapter 6 and 7 demonstrated the potentialities of coupling in vivo nose-space analysis with dynamic sensory analysis to better investigate the complex relation between aroma release and perception and the impact of product and consumer characteristic on this relation. In Chapter 6, the nose-space PTR-QiToF- Appendix 226 Summary 226 MS approach allowed to conclude that when consuming composite foods, decrease in sensory intensity perception was not due to a lower delivery of aroma compounds into the nasal cavity, as in-nose aroma release of condiments increased with the presence of a carrier food. Consequently, cognitive effects are likely to play a key role in sensory perception of composite foods. In a global context where agroindustry should contribute to more healthy and sustainable diets by reducing salt, sugars and fats levels in their processed food products, in vivo nose-space measurement coupled with sensory analysis is a valuable method for better predicting product reformulations effects. In Chapter 7, the same approach gave indications that consumers origin may affect aroma release and perception of mint chewing gum. Differences in aroma release could be due to many different factors. Participant’s origin (and diets) may be responsible for physiological differences especially in terms of saliva composition and oral/nasal microbiota. If these observations would be validated by larger population studies, they would be of great interest for the new emerging field of personalized product design and nutrition. Conclusion: This thesis, investigated several applications of PTR-MS in the context of food quality at agroindustry level with a focus on the sensory dimension of aroma. When looking at food quality control programs, high throughput VOCs fingerprinting of food samples through PTR-ToF-MS coupled with a multipurpose autosampler, represents a promising approach to select raw materials that will need quality sensory evaluations. The same approach is suitable for monitoring changes during shelf life at different conditions and, more in general for on-line applications during cooking operations. In vivo nose-space measurements coupled to sensory dynamic methods is a valid approach to investigate impact of product and consumer characteristics on aroma release and flavour perception during food consumption. In conclusion, PTR-MS approach was proofed to be one example of technological application, which can greatly facilitate the understanding of the factors affecting food quality from agricultural raw materials to market food products