Our diet plays a key role in our health, and a vast amount of research has been dedicated to the relationship between certain dietary patterns and health outcomes like disease. Traditional methods for collecting data on food intake, like food diaries and food frequency questionnaires, can often be imprecise and inaccurate, and are especially challenging for some patient populations such as people with Alzheimer’s disease. A recent advancement in the area of untargeted metabolomics, developed by an international research team led by University of California San Diego scientists, has the potential to improve research into dietary patterns by allowing food intake to be empirically assessed using data from biospecimens.
Current metabolomics studies annotate or identify only 10% of molecular features in sampled specimens, leaving 90% of material unknown. The new approach uses reference-data-driven (RDD) analysis to match metabolomics data derived from tandem mass spectrometry (MS/MS) against metadata-annotated data as a pseudo-MS/MS reference library. This increases MS/MS spectra usage over conventional structural MS/MS library matches, the authors wrote. To demonstrate how this approach could be used for untargeted metabolomics related to diet, the team created a food metabolomics reference dataset including untargeted metabolomics and detailed and structured metadata from around 3,500 foods. The dataset includes 157 different food-specific metadata fields and contains nearly 108,000 unique MS/MS spectra merged from a total of more than 1.9 million spectra.
The researchers used information from controlled research diets of participants in a sleep and circadian study to determine if RDD could recover foods known to be consumed. The analysis detected 11 categories directly matched to foods provided in the controlled diet, and four mismatched categories, three of which could be explained – for example, rosemary, which is a common ingredient added to ground meat but not always written on package ingredient lists. The test demonstrated that RDD could successfully obtain correct diet information from untargeted metabolomics data, as well as to monitor diet adherence in controlled-diet studies, the authors wrote. Additionally, the researchers used RDD to parse dietary patterns, such as vegan and omnivore, and the team found large improvements of how many molecules in blood and stool could be explained when food items were matched to population, such as matching food from Italy to people from the Cilento peninsula, where UC San Diego scientists are collaborating on a study of centenarians. This research was published in Nature Biotechnology.
“This study also points to a way toward using RDD to explain the dark matter in our metabolome, not only in terms of diet, but in exposures to chemicals from the clothes we wear, the medications we take, the beauty products we apply and the environments we are exposed to,” said co-corresponding author Pieter Dorrestein, director of the Collaborative Mass Spectrometry Innovation Center in the Skaggs School of Pharmacy and Pharmaceutical Scientists at UC San Diego. “It will truly let us explore the chemical connections between ourselves and the world we inhabit.”
Overall, the RDD approach to untargeted metabolomics improved data output more than five-fold over conventional techniques. Co-corresponding author Rob Knight, director of the Center for Microbiome Innovation at UC San Diego, noted the technique could also be used to assess dietary information of animals in wildlife conservation applications.