Maternal and neonatal metabolomes and their associations to immune maturation and allergy in early life

Abstract: Allergy, one of the most common chronic diseases worldwide, is caused by a dysregulated immune system reacting to normally harmless proteins. However, regulating mechanisms are not well understood. The aim of this thesis was to investigate if plasma and placenta metabolites associate prospectively to allergy development and immune maturation. The aim was further to explore differences between arterial and venous umbilical cord blood metabolomes, and if they associated with maternal or infant traits. Placentas and plasma (maternal from pregnancy and delivery and from the umbilical cord) were obtained from the prospective NICE-cohort. Metabolites were measured by LC-MS and GC-MS. None of the measured metabolomes associated with any of the investigated allergic outcomes (asthma, food allergy and eczema). Modest associations were observed between immune maturation (in particular memory B cells) and plasma and placenta metabolomes. Energy-related metabolites were higher in arterial cord blood, while amino acids were higher in venous cord blood. Amino acid and energy metabolites were higher in first born children compared to children with older siblings. Overall, the results suggest that immunomodulatory metabolites might be transferred from mother to child during pregnancy, affecting the future production and maturation of immune cells. Studies involving umbilical cord should consider the differences in arterial and venous cord blood and the association of maternal parity in experimental design and data analysis. Furthermore, algorithms for real-time quality monitoring in untargeted LC-MS metabolomics were developed to improve quality during data generation. Quality monitoring was based on general metrics (e.g. total intensity and number of peaks) and peak metrics from so-called landmark features (e.g. peak area and noise). Landmark features were discovered, validated and then extracted and used in procedures to automatically discover injections of poor data quality. The developed procedures show great promise for improved data generation in high-throughput metabolomics.