부산대학교 도서관

Background. Paediatric studies examining cardio-metabolic risk indicators use small for gestational age (SGA) as a surrogate marker for fetal growth restriction (FGR). However, FGR can exist without SGA. Hypothesis. Antenatal markers of higher FGR risk can identify fetuses at greatest risk of long-term cardio-metabolic disease. Methods. i) Children aged three to six years attended for measurements indicating glucose metabolism, lipid metabolism and vascular health status. The majority were born following pregnancies with higher FGR risk identified. Alongside traditional statistical methods, Random Forest Classification (RFC) was used to test the hypothesis. ii) Following pregnancies with higher FGR risk, infants had auxological measurements at birth, three, six and 12 months. Infant weight and adiposity trajectories were determined, to examine relationships with fetal weight trajectory. iii) 'Omic data from a subset of children were used to determine differentially expressed genes and metabolites between quartiles of weight trajectory. Pathways analysis was undertaken. K-means clustering was also adopted, as an unsupervised method. Rank regression with fetal, and then childhood weight trajectory as the dependent variable, enabled construction of hypernetworks. Selecting highly correlated genes within the hypernetworks, RFC models were created to predict the highest quartile of SBP. Results. i) Significant correlations linking antenatal, fetal and childhood measurements were established. A triad of correlations was found between fetal, childhood weight trajectory and systolic blood pressure (SBP). Regression analyses supported the role of childhood weight trajectory as a potential mediator. RFC demonstrated that antenatal markers could predict highest quartile of childhood indicators. ii) Negative correlations were established between fetal weight trajectory and birth to six-month adiposity trajectories, but not with infant weight trajectory. iii) Ornithine was differentially expressed across all supervised analyses, implicating the urea cycle. Elucidation of ARG1, encoding arginase (which catalyses the hydrolysis of arginine to ornithine) supported this. Unsupervised analysis revealed two participant clusters exhibiting SBP differences. We discovered that this separation can be largely accounted for by expression of 47 birthweight-related genes from a GWAS meta-analysis. Hypernetworks including significant genes from rank regression, allowed identification of sets of functionally related genes. Both predicted the highest quartile of SBP in RFC models. Conclusions. Fetal weight trajectory can be used to identify fetuses who may develop higher SBP. This could be driven through childhood weight trajectory. Infant adiposity trajectories may aid early-life detection of at-risk individuals. Longer-term follow up of both cohorts is required to establish impact on disease risk.