Genetic and non-genetic influences on birth weight and type 2 diabetes

Thesis

Type 2 diabetes (T2D) is a complex condition that is characterised by hyperglycaemia and associated with over 400 independent genetic variants to date, as well as non-genetic factors. Birth weight, an indicator of intrauterine nutrition and predictor of infant survival, is also associated with the risk of T2D in adulthood. In multiple study populations, the association between birth weight and the risk of subsequent T2D forms a U-shaped curve, with lower and higher birth weights associated with greater risk of T2D. This complex relationship between birth weight and subsequent risk of T2D is influenced by an interplay of genetic and non-genetic influences.

To explore the genetics of birth weight and T2D and to investigate the relationship between these traits, I conducted genome-wide association studies (GWAS) of birth weight. Seven different models of GWAS of birth weight were conducted, using various combinations of covariates, imputed, or directly genotyped variants, and maternal and foetal genotypes. None of the GWAS models yielded genome-wide significant (p < 5.0 × 10^-8) associations. Certain signals that have high levels of significance across multiple models could still merit further exploration.

I also quantified the relationship between polygenic scores (PS) that were constructed using clusters of T2D-associated genetic variants and T2D and related phenotypes. These analyses were based on data that had been collected in a longitudinal study of diabetes, which was conducted in an Indigenous study population from the Southwestern US between 1965 and 2007 and had data for birth weight, T2D, T2D subphenotypes, population substructure, and genotype. Partitioned/process-specific polygenic scores (pPS) were constructed using genotypes from the longitudinal study population and genetic clusters that were identified by Mahajan et al to be associated with T2D subphenotypes. PS were constructed using genotypes from the longitudinal study population and summary statistics from meta-analyses of GWAS of T2D. Results broadly agree with putative pathophysiological mechanisms that underlie each genetic cluster by Mahajan et al.

I also investigated the prediction of the incidence of T2D using clinical and genetic variables in the longitudinal study. The performances of predictive models including various combinations of T2D-related clinical variables and T2D PS were assessed and compared. All T2D PS were significantly associated with T2D and, when added to models including clinical variables alone, exhibited a significant but small improvement in the prediction of T2D incidence.

This suggests that further improvements in understanding the genetics of T2D and related traits can result in improved understanding of diabetes pathophysiology and clinical applications of PS and pPS.

Authors: Wedekind, LE

• DOI: 10.5287/ora-zbzo7yrkn
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: type 2 diabetes; polygenic scores; epidemiology; GWAS