The Impact of Variability in Body Mass Index on Cardiovascular Health

Background: Over 529 million individuals live with diabetes, with approximately 96% affected by type 2 diabetes (T2D), a condition partly driven by rising global obesity rates. Weight loss therapies are central to T2D management. There are often significant fluctuations in body mass index (BMI). While some evidence links BMI variability with increased cardiovascular disease (CVD) risk, this is a controversial topic that lacks consensus. Aim: This thesis aims to investigate the relationship between BMI variability and CVD, examine the genetic determinants of BMI variability, and assess its utility in enhancing clinical risk prediction. Methods: A systematic review and meta-analysis was performed to aggregate data from existing literature on BMI and weight fluctuations and CVD, establishing a summative risk estimate. The robustness of this association was further evaluated through linear, logistic, and Cox proportional hazards regression analyses across three clinical trial cohorts and one observational cohort. An extended Cox model incorporating time-dependent covariates was also applied within the Tayside Bioresource cohort. Additionally, five genome-wide association studies (GWAS) were performed across separate cohorts to investigate the relationship between BMI variability and genetic variants (minor allele frequency (MAF) >0.01). Meta-analysis of GWAS summary statistics facilitated the estimation of heritability and the development of a polygenic risk score (PRS) for BMI variability. Finally, the PRS was integrated with QRISK3 using Cox regression and receiver operating characteristics to evaluate its predictive enhancement for CVD. Results: The systematic review and meta-analysis indicated that increased BMI variability is associated with elevated CVD risk, despite significant heterogeneity among studies. This association was robustly replicated across multiple cohorts and sustained under time-dependent modelling. Meta-GWAS identified 32 loci linked to BMI variability, though the derived heritability of less than 1% implies low genetic influence. Importantly, incorporating the BMI variability PRS modestly but significantly improved the predictive performance of the QRISK3 model. Conclusion: This thesis demonstrates that BMI variability is a robust, independent predictor of CVD risk, with genetics playing a minor role, and highlights the potential clinical utility of integrating a BMI variability PRS into cardiovascular risk assessment frameworks.