Introduction/Background: Guidelines from the World Health Organization currently recommend targeting a body mass index (BMI) between 18.5 and 24.9 kg/m2 based on the lowest risk of mortality observed in epidemiological studies. However, these recommendations are based on population observations and do not take into account potential inter-individual differences. We hypothesized that genetic and non-genetic differences in adiposity, anthropometric, and metabolic measures result in inter-individual variation in the optimal BMI. Methods: Genetic variants associated with BMI as well as related adiposity, anthropometric, and metabolic phenotypes (e.g. triglyceride (TG)) were combined into polygenic risk scores (PRS), cumulative risk scores derived from the weighted contributions of each variant. 387,692 participants in the UK Biobank were split by quantiles of PRS or clinical biomarkers such as C-reactive protein (CRP), and alanine aminotransferase (ALT). The BMI linked with the lowest risk of all-cause and cause-specific mortality outcomes (“nadir value”) was then compared across quantiles (“Cox meta-regression model”). Our results were replicated using the non-linear mendelian randomization (NLMR) model to assess causality. Results: The nadir value for the BMI–all-cause mortality relationship differed across percentiles of BMI PRS, suggesting inter-individual variation in optimal BMI based on genetics (p = 0.005). There was a difference of 1.90 kg/m2 in predicted optimal BMI between individuals in the top and bottom 5th BMI PRS percentile. Individuals having above and below median TG (p = 1.29×10-4), CRP (p = 7.92 × 10-5), and ALT (p = 2.70 × 10-8) levels differed in nadir for this relationship. There was no difference in the computed nadir between the Cox meta-regression or NLMR models (p = 0.102). Conclusions: The impact of BMI on mortality is heterogenous due to individual genetic and clinical biomarker level differences. Although we cannot confirm that are results are causal, genetics and clinical biomarkers have potential use for making more tailored BMI recommendations for patients. / Thesis / Master of Science (MSc) / The World Health Organization (WHO) recommends targeting a body mass index (BMI) between 18.5 - 24.9 kg/m2 for optimal health. However, this recommendation does not take into account individual differences in genetics or biology. Our project aimed to determine whether the optimal BMI, or the BMI associated with the lowest risk of mortality, varies due to genetic or biological variation. Analyses were conducted across 387,692 individuals. We divided participants into groups according to genetic risk for obesity or clinical biomarker profile. Our results show that the optimal BMI varies according to genetic or biomarker profile. WHO recommendations do not account for this variation, as the optimal BMI can fall under the normal 18.5 - 24.9 kg/m2 or overweight 25.0 – 29.0 kg/m2 WHO BMI categories depending on individual genetic or biomarker profile. Thus, there is potential for using genetic and/or biomarker profiles to make more precise BMI recommendations for patients.

Modern medical practice and science involve complex phenotypic definitions. Understanding patterns of association across this range of phenotypes requires co-analysis of high-dimensional association studies in order to characterise shared and distinct elements. In this thesis I address several problems in this area, with a general linking aim of making more efficient use of available data. The main application of these methods is in the analysis of genome-wide association studies (GWAS) and similar studies. Firstly, I developed methodology for a Bayesian conditional false discovery rate (cFDR) for levering GWAS results using summary statistics from a related disease. I extended an existing method to enable a shared control design, increasing power and applicability, and developed an approximate bound on false-discovery rate (FDR) for the procedure. Using the new method I identified several new variant-disease associations. I then developed a second application of shared control design in the context of study replication, enabling improvement in power at the cost of changing the spectrum of sensitivity to systematic errors in study cohorts. This has application in studies on rare diseases or in between-case analyses. I then developed a method for partially characterising heterogeneity within a disease by modelling the bivariate distribution of case-control and within-case effect sizes. Using an adaptation of a likelihood-ratio test, this allows an assessment to be made of whether disease heterogeneity corresponds to differences in disease pathology. I applied this method to a range of simulated and real datasets, enabling insight into the cause of heterogeneity in autoantibody positivity in type 1 diabetes (T1D). Finally, I investigated the relation of subtypes of juvenile idiopathic arthritis (JIA) to adult diseases, using modified genetic risk scores and linear discriminants in a penalised regression framework. The contribution of this thesis is in a range of methodological developments in the analysis of high-dimensional association study comparison. Methods such as these will have wide application in the analysis of GWAS and similar areas, particularly in the development of stratified medicine.

Background: Coronary Artery Disease (CAD) represents the leading cause of mortality and morbidity worldwide despite declines in the prevalence of environmental risk factors. This trend has drawn attention to the risk conferred by genetic variation. Twin and linkage studies demonstrate a profound hereditary risk for CAD, especially in young individuals. Rare genetic variants conferring high risk for extreme disease phenotypes can provide invaluable insight into novel mechanisms underlying CAD development. Methods: Whole exome sequencing was performed to characterize rare protein-altering variants in 52 early-onset CAD (EOCAD) patients encompassing the DECODE study. The enrichment of Mendelian dyslipidemias in EOCAD was assessed through interrogation of pathogenic mutations among known lipid genes. The identification of novel genetic CAD associations was conducted through case-only and case-control approaches across all protein-coding genes using rare variant burden and variance component tests. Lastly, beta coefficients for significant risk genes from the European population in the Early-onset Myocardial Infarction (EOMI) cohort (N=552) were used to construct calibrated, single-sample rare variant gene scores (RVGS) in DECODE Europeans (N=39) and a local European CAD-free cohort (N=77). Results: A 20-fold enrichment of Familial hypercholesterolemia mutation carriers was detected in EOCAD cases compared to CAD-free controls (P=0.005). Association analysis using EOMI Europeans revealed exome-wide and nominal significance for two known CAD/MI genes: CELSR2 (P=1.1x10-17) and APOA5 (P=0.001). DECODE association revealed exome-wide and nominal significance for genes involved in endothelial integrity and immune cell activity. RVGS based upon beta coefficients of significant CAD/MI risk genes were significantly increased in DECODE (z-score=1.84; p=0.03) and insignificantly decreased among CAD-free individuals (z-score=-1.61; p=0.053). Conclusion: Rare variants play a pivotal role in the development early CAD through Mendelian and polygenic mechanisms. Construction of RVGS that are calibrated against population and technical biases can facilitate discovery of single-sample and cohort-based associations beyond what is detectable using standard methods. / Thesis / Master of Science (MSc)