GC-MS based metabolomics – Development of a next generation GC-MS method and its application to nutrition and biomarker research
Metabolomics, the measurement of a broad range of small molecules in a sample, is maturing as a field in analytical chemistry and becoming a standard research tool in biological sciences for helping to generate a wider understanding of the complex biological mechanisms behind, for example, development of disease and effects of diet on health. Metabolomics analysis is generally divided into either targeted or untargeted metabolomics. The former ‘targets’ a specific set of molecules, often quantitatively, while the latter aims to detect as many metabolites as possible in the sample. Both of these approaches have their advantages and disadvantages, and combining both into one single analytical method could allow the specificity, sensitivity, and quantitation of targeted metabolomics to be combined with the broad coverage and potential to find unknown compounds that are the key advantages of untargeted metabolomics.
In this thesis, a novel gas chromatography triple quadrupole metabolomics (GC-MS) method that exploits fast data acquisition to simultaneously acquire both targeted quantitative data using multiple reaction and selected ion monitoring, and untargeted qualitative data using full spectrum scanning. The method was developed for human blood plasma and applied to two studies. The first was a crossover intervention study comparing metabolic effects of consumption of herring with chicken and pork in 15 adults. The second was a prospective cohort of 600 older Swedish women for discovering predictive biomarkers of development of type 2 diabetes (T2D) and for exploration of the associations between potential dietary and nutrient biomarkers and glucose tolerance status and development of T2D in the same cohort.
The evaluated method parameters (linearity, limit of detection and quantification, precision, accuracy, number of spectral features) supports the approach of combining targeted quantitative and untargeted qualitative data acquisition as a way to improve GC-MS metabolomics. The method was successfully applied to the analysis of 1200 samples from 600 subjects, measured over 24 separate analytical batches with an average within batch metabolite variation of 10 %, based on control sample metabolites detected using multiple reaction monitoring.
In the intervention study, the new metabolomics method detected 190 identified metabolites, of which 18 were altered when subjects ate herring instead of chicken and pork. These changes were mainly around the tricarboxylic acid and urea cycles, with an apparent differential effect related to arginine metabolism. This finding was supported by finding that circulating nitric oxide was higher in male subjects after the herring diet compared to the chicken and pork diets. In the cohort study, we established that the best predictive markers of T2D detected using metabolomics improved on or had similar prediction to established predictors of T2D. Using a combination of both the targeted and untargeted data, we were also able to detect 10 dietary and nutrient biomarkers, many of which were strongly associated with glucose tolerance status and development of T2D in this cohort. This is one of the first studies using multiple dietary and nutrient biomarkers that suggests a clear role of diet in prevention of T2D. This supports the current dietary guidelines for eating whole grains and fish for preventing T2D.
In conclusion, the metabolomics method developed as part of this thesis detects a wide array of known biomarkers in blood plasma while still providing global untargeted information and having the possibility of expansion by addition of targeted molecules of interest. The method proved to be robust during the analysis of a moderately sized sample set, supporting its further use in observational cohorts. Results from the application of this metabolomics method further support the potential of metabolomics to add value to biological research by highlighting diverse outcome- metabolite relationships that may otherwise be overlooked.
type 2 diabetes