Despite the rapid increase in both adiposity and diabetes in China, substantial uncertainty remains about the relationship between these two conditions in the population. Using data from the China Kadoorie Biobank study of 0.5 million adults recruited during 2004-8 from 10 diverse areas of China, this thesis examines the associations of different adiposity measures (overall adiposity: BMI and percentage body fat; central adiposity: waist circumference, waist-hip ratio, and waist-height ratio) with type 2 diabetes prevalence and incidence. To assess the quality of diagnosis, a separate event-verification study was conducted in ~1,000 reported diabetes cases. Overall at baseline, the mean age of the analysed participants was 52 years, 41% were men, 32% had a BMI≥25 kg/m² (4% ≥30 kg/m²) and 5.2% had self-reported or screen-detected diabetes. Both cross-sectional and prospective analyses of well-characterised diabetic cases (26,622 prevalent and 2,910 incident cases) showed that adiposity is strongly positively associated with diabetes (p<0.0001), throughout all or most of the distribution of each adiposity measure. Per 1 SD higher adiposity measure, measures of central adiposity were associated with ~90% increased risk, compared with ~80-85% increased risk for general adiposity measures. Among measures of central adiposity, waist-hip ratio was the most strongly associated with diabetes prevalence, whereas waist circumference was the most predictive of diabetes incidence. Although measures of central adiposity were the most strongly associated with diabetes risk, there was still a strong positive association with measures of general adiposity after adjusting for central adiposity (p<0.0001), and the combination of both types of measure improved risk prediction. Given waist circumference, hip circumference was inversely associated with both diabetes prevalence and incidence (p<0.0001). For many of the above associations, there was possible effect modification by age and sex. These findings will provide important and reliable evidence to quantify the level of diabetic risk associated with adiposity, hence to inform clinical interventional strategies and future public health programmes.

In recent times the potential impact of preeclampsia on the cardiovascular health of offspring has been identified. This thesis explores the relationship between preeclampsia and offspring cardiovascular phenotype during the first three decades of life. A systematic review and meta-analysis provided evidence that there was increased blood pressure and BMI in the offspring of preeclamptic pregnancies (n = 45,249). There was however limited data on metabolic features and inadequate characterisation of the degree of prematurity or growth restriction in existing literature. I therefore studied data on two birth cohorts with up to 28 years of detailed prospective follow up (n = 2868 and n = 926). Those born very preterm to preeclamptic pregnancies had transient perinatal reductions in insulin and cholesterol, although extreme prematurity was the only determinant of variation in cardiovascular risk in later life, with changes in both metabolism and blood pressure. In those born closer to, or at term, gestation was no longer relevant and an independent impact of preeclampsia on blood pressure was evident, so that by age 20, those born at term to preeclamptic pregnancies were four and a half times more likely to demonstrate clinically-apparent hypertension. I then investigated whether there were changes in other features of cardiovascular phenotype, independent of blood pressure, in preterm neonates born following preeclampsia (n = 46). At 3 months of age preterm infants born to hypertensive pregnancies had subclinical alterations in cardiac strain, independent of gestation or birth weight but not differences in blood pressure, or microvascular structure. These findings highlight preeclampsia and prematurity as key, independent perinatal factors, important in determining cardiovascular phenotype and risk during early life. Preeclampsia is associated with a specific lean, hypertensive phenotype, associated with cardiac functional alterations; these findings begin to define a distinct at risk population who may require targeted preventative interventions.