UCLA

Tobacco smoke contains multiple toxic compounds, and maternal tobacco smoking during pregnancy has been related to negative infant and child outcomes. Newborn blood spots are collected shortly after birth to test the baby for a set of metabolic disorders, after which the State of California stores them for research use. High-resolution metabolomics (HRM) is an analytical approach utilizing ultra-high resolution mass spectrometry and data science methods to characterize and quantify small molecules (metabolites) in biological samples. In this dissertation, we employed HRM as a tool to demonstrate the usefulness of neonatal dried blood spots (DBS) and nicotine metabolites in epidemiological studies, and to better understand the biological pathways through which maternal tobacco use may have long-term impacts on child metabolism.We first examined the utility of archived newborn blood spots which have been stored for 29 years. We used 899 neonatal DBS of children without cancer before age 6. High-resolution metabolomics with liquid chromatography mass spectrometry (LC-MS) was performed and the relative ion intensities of common metabolites and selected xenobiotic metabolites of nicotine (cotinine and hydroxycotinine) were evaluated. In total, we detected 26,235 mass spectral features across two separate chromatography methods (C18 and HILIC). For most of the 39 metabolites related to nutrition and health status, we found no statistically significant annual trends across the years of storage. Nicotine metabolites were captured in the DBS with relatively stable intensities. We then assessed the usefulness of nicotine biomarkers in the same population and built a prediction model for maternal tobacco smoking in pregnancy based on birth certificate information using a combination of self- or provider-reported smoking and biomarkers (smoking metabolites cotinine and hydroxycotinine) in neonatal blood spots as the alloyed gold standard. Potential predictors of smoking were selected from the California birth certificate. Logistic regression with stepwise backward selection was used for prediction model building. Five predictors were selected by the stepwise procedure, including maternal race/ethnicity, maternal education, child’s birth year, parity, and child’s birth weight. We calculated an overall discrimination accuracy of 0.724 and an AUC of 0.805 (0.770-0.839) in the training set. Similar accuracies were achieved in the internal and external validation sets. Lastly, we performed a HRM analysis in 899 newborns, following an untargeted metabolome-wide association study (MWAS) workflow. A total of 26,183 features (15,562 in HILIC column and 10621 in C18 column) were detected with HRM of which 1,003 were found to be associated with maternal smoking. Smoking affected metabolites and metabolic pathways in neonatal blood included vitamin A (retinol) metabolism, the kynurenine pathway, and tryptophan and arachidonic acid metabolism. The metabolites and pathway perturbations associated with cigarette smoking that we identified suggested inflammatory responses and have also been implicated in chronic diseases of the central nervous system and the lung. In summary, our studies support the usefulness of DBS stored long-term for epidemiological studies of the metabolome, build a prediction model that may benefit future birth registry-based studies in California when there is missing maternal smoking information, and suggest that infant metabolism in the early postnatal period reflects smoking specific physiologic responses to maternal smoking with strong biologic plausibility.