Human Milk Components Are Associated With Gene-Environment Interactions and Impact Childhood Respiratory and Atopic Health

Abstract

Human milk (HM) contains a myriad of components, including oligosaccharides (HMOs), fatty acids (HMFAs), and microbiota (HMM), produced at variable concentrations among lactating mothers. Despite the well-established health benefits of HM, the factors shaping HM composition and the mechanisms by which human milk components (HMCs) impact the lung and atopic health of HM-fed infants remain poorly understood. This dissertation leverages data from HM samples collected from 1,206 mothers in the CHILD Cohort Study at 3-6 months postpartum. In CHILD, genomic DNA was extracted from blood samples of 2,552 mothers and 2,967 infants, and childhood health outcomes were assessed from birth to age 5 years. An analytical framework, progressing from univariate analysis to gene-environment interaction (G×E), network-based modelling, and supervised machine learning, was applied to investigate the genomic and environmental determinants of HMCs and how these components, individually and in combination, influence childhood asthma and atopy. In the first paper, G×E analyses of HMOs and childhood health outcomes identified single nucleotide polymorphisms (SNPs) not detected in previous genome-wide association studies (GWASs). Specifically, we reported maternal SNPs associated with HMOs in certain environmental contexts, as well as infant SNPs associated with childhood respiratory and atopic outcomes following exposure to specific HMOs. The second paper presents the first GWAS of HMM, showing that maternal genomics contributes to shaping HMM composition. We identified network clusters of co-occurring microbes, such as a Yellow cluster that included microbes Veillonella and Prevotella, which was associated with reduced childhood atopy. The third paper integrates HMOs, HMFAs, and HMM with infant polygenic risk scores (PRS) to assess their combined impact on childhood asthma and atopy. Our G×E analyses determined interaction effects between HMCs and infant PRS on childhood atopy. A machine learning model incorporating all HMCs and infant PRS significantly outperformed models using any individual HMC or infant PRS alone in predicting asthma and atopy risk. By integrating HMCs with host genomics using diverse computational methods, this dissertation highlights a data-integration framework for studying the effects of HMCs on child health and supports the development of targeted therapeutic interventions, such as prebiotics and probiotics, for asthma and atopy prevention.