Seaweed, Spinach, and Coffee – Discovery and Validation of Novel Biomarkers of Food Intake by Untargeted Metabolomics

Research output: Book/ReportPh.D. thesisResearch

  • Muyao Xi
Dietary assessment approaches play a major role in nutrition research and might be used more to explore the associations between foods and their effects on health. The information obtained from the traditional assessment of dietary intake is inaccurate and subjective due to recall bias, limited food lists, etc. Therefore, food intake is often either over- or under-estimated, complicating the establishment of cause and effect relationships between food and health and, consequently, the formulation of public health recommendations. Data-driven untargeted metabolomics, employing advanced analytical technologies, allows to measure thousands of metabolites simultaneously, thereby supporting the measurement of metabolites deriving from the intake of specific foods in
biofluids. These specific compounds, identified and validated as biomarkers of food intake (BFIs), can serve as an alternative to traditional methods of assessing food consumption or can complement these in evaluating food-health associations adequately.
Despite plant-based foods having well-known nutritional and potential bioactive properties, they lack specific BFIs which may be crucial for the objective assessment of their effects on health. This PhD thesis focuses on the identification and validation of BFIs of three consumed plant-based foods from ocean to land, i.e., seaweed, spinach, and coffee. This PhD thesis begins with a comprehensive systematic review to collate previously reported specific compounds associated with seaweed consumption (paper Ⅰ). Several candidate BFIs related to seaweed were retrieved from human trials published in the literature. The reported candidate BFIs needed further validation, and additional work was necessary to verify their usefulness as dietary assessment tools. Therefore, a three-way crossover human intervention trial was carried out, in which meals with two kinds of brown seaweed and a control meal were served in a randomized order to participants (paper Ⅱ). Biological fluids were collected and analyzed through an untargeted metabolomics approach, by using ultra-high-performance liquid chromatography-quadrupole coupled to time-of-flight mass spectrometry (UHPLC-QTOF-MS). We discovered four seaweed-related BFIs. Whereas one was identified at level Ⅰ, i.e., matching with an analytical standard in terms of RT, m/z, and spectral pattern, the remaining three need further thorough structural elucidation. Consideration of combined BFIs in future research might solve the issue that none of these BFIs were specific individually to brown seaweed.
The third work of the thesis regards the identification of spinach intake biomarkers by applying an untargeted metabolomics approach on a cross-over human intervention study (paper Ⅲ). However, even though the samples were collected from a cross-over study with whole leafy and minced spinach, we analyzed the study as a simple before-and-after design due to the lack of a proper control group (with no spinach intake). Three candidate BFIs were found, and 69 effect-like biomarkers, i.e., exhibiting altered levels in urine after spinach intervention, were selected by a combination of PLSDA and student’s t-test. All markers were unaffected by spinach processing. The three candidate BFIs are des-amino arginine pentenol ester, a malic acid ester of p-coumarate, and an unidentified isomer of the latter. Most of the effect-like biomarkers were associated with meal structure instead of spinach intake per se. Nevertheless, the discovery of actual effect biomarkers is a promising option in the future, as the combination of BFIs and effect biomarkers could be a relatively comprehensive tool to objectively measure the food intake, as well as the corresponding metabolic influence in humans.
In general, the discovery of BFIs is only the first part of biomarkers development. Subsequent validation of the BFIs based on published criteria is the last and essential step. The validation of BFIs consists in evaluating whether they are capable of predicting targeted food intake in another independent, less-controlled study. The validity of the BFIs of both spinach and seaweed intake
discussed in this thesis needs to be confirmed in an independent study in the future.
The fourth work of the thesis consisted of identifying and validating coffee intake biomarkers by applying an untargeted metabolomics approach on four independent human studies (paper Ⅳ). Through ASCA and PLS-DA statistical analysis, 23 putative urinary biomarkers strongly associated with coffee consumption were selected in common in three parallel intervention studies and identified. Sixteen of the 23 markers were validated against food diaries in the cross-sectional NU-AGE study. Several validation criteria, including plausibility, robustness, time-response relationship, and reproducibility of these 16 candidate BFIs were confirmed. Their dose-response relationship was also
partially validated in the NU-AGE study, and this characteristic needs to be fully validated in an intervention study with several doses of coffee. Due to the non-specificity of some of these BFIs to coffee intake, combinations of BFIs were evaluated. The most parsimonious set for total consumers and high consumers gave a perfect prediction of coffee consumption (AUCtest = 1 and ERtest = 0).
1H-pyrrole-2-carboxaldehyde sulfate and 3’,4’- dihydroxyacetophenone (DHAP) sulfate was the most parsimonious set for total consumers; ferulic acid sulphate and 4-ethylcatechol glucuronide (GlcA) was the most parsimonious set for high consumers. For low consumers, the most parsimonious set consisted of three candidate BFIs (1H-pyrrole-2-carboxaldehyde sulfate, ferulic acid sulphate, and
3’,4’-DHAP sulfate) and had a very good prediction performance (AUCtest = 1, ERtest = 0.0385).
Validation of BFIs specific to spinach or seaweed intake remains a subject for future research. The parsimonious sets of coffee intake biomarkers allowed the assessment of coffee intake levels in human studies, but a dose-response intervention is still needed to validate the ability of BFIs to quantify coffee intake accurately. The robustness and reproducibility of combined BFIs associated with coffee intake deserve further validation.
Original languageEnglish
Place of PublicationCopenhagen
PublisherDepartment of Nutrition, Exercise and Sports, Faculty of Science, University of Copenhagen
Number of pages118
ISBN (Print)9788772094144
Publication statusPublished - 2021

ID: 272410586