Combining Mendelian Randomization and Randomized Control Trial Study Designs to Determine Effects of Adiposity on the Plasma Proteome

Mendelian randomization (MR) studies using a UK blood donor cohort (INTERVAL) have revealed that higher BMI has a substantial effect on the plasma proteome. We aimed to determine the effect of adiposity on the circulating proteome by combining MR and a randomized control trial (RCT). We used SomaLogic proteomic data from the Diabetes Remission Clinical Trial (DiRECT). Participants with obesity and type 2 diabetes were allocated either guideline-based care (control) or total diet replacement (TDR) (intervention) treatment. Serum samples were taken at baseline and after 12 months. Participants in the TDR group had mean BMI change of -3.6kg/m2 vs -0.46kg/m2 in the control group. We performed linear regression to explore the association between BMI change and protein change. To reduce confounding and utilize RCT study design, treatment group was used as an instrument for BMI change in a two stage least squares analysis. INTERVAL analyses suggested a causal effect of BMI on proteins including fumarylacetoacetase (0.51 SDs higher per SD higher BMI, 95% CI 0.30-0.72) and growth hormone receptor (GHR) (0.43 SDs, 95% CI 0.23-0.63). The DiRECT study provided evidence that proteins that are positively associated with BMI can be reduced by lowering BMI: 1 SD reduction in BMI reduces fumarylacetoacetase levels by 0.73 SDs (95% CI 0.55-0.90) and the GHR by 0.57 SDs (95% CI 0.42-0.73). Concordance across study designs, each with different potential sources of bias, gives increased confidence in the estimated causal effect of BMI on selected proteins, pointing to their potential involvement in downstream health outcomes.