A bidirectional causal association between type 2 diabetes and hypertension based on the life course approach

Abstract

Background: The positive relationship between type 2 diabetes (T2D) and hypertension has been proved in a number of large observational studies. These observational studies were limited in confirming causal relationships because of the potential confounding biases and reverse causality. There was only one previous Mendelian randomization (MR) study for the bidirectional causal relationship between T2D and hypertension was conducted, it had ethnical limitations. This study aimed to conduct improved MR study in a Korean population-based longitudinal cohort study and investigate the bidirectional causal relations of fasting blood sugar (FBS) levels with systolic and diastolic blood pressure (BP) using MR analysis, and validate the bidirectional causal association based on life course approach. Methods: Five MR methods were applied, including the two-stage least squares (2SLS) regression method, inverse-variance weighted (IVW) method, and 2 median-based methods (simple and weighted), MR-Egger was used to assess the bidirectional causal association. The weighted genetic risk score (wGRS) for genetically instrumented FBS and systolic blood pressure (SBP) was constructed using 91 and 68 single nucleotide polymorphisms (SNP) extracted from the GWAS of the large Korean biobank. The p-value cutoff was set at <1.0×10-8 based on multiple linear regression. A trajectory analyses was performed to estimate how much genetically determined FBS or SBP value estimated from IVs affects future T2D or hypertension incidents. the Cox proportional hazard models were conducted to assess the association analyses between wGRS and future T2D or hypertension in a general healthy population. To evaluate the association analyses between trajectories for genetically determined FBS or SBP value and future T2D or hypertension in a general healthy population, the Cox proportional hazard models were performed. Results: MR analysis using the two-stage least squares regression method adjusted for age and sex showed that FBS elevation by 10 mg/dL due to our genetic variants was associated with an increased SBP of 1.63 mm/Hg (p=0.005), and genetically determined elevation of SBP by 10 mm/Hg was associated with an FBS increase of 11.39 mg/dL (p<0.0001). Using the MR-Egger method, when the FBS was 1 mg/dL higher genetically, it was associated with a higher SBP of 0.20 mm/Hg (p=0.005, p for intercept=0.823). Meanwhile, an elevated SBP of 1 mm/Hg genetically was associated with an increased FBS of 1.08 mg/dL, and a significant intercept p-value was demonstrated (p<.0001 , p for intercept=0.001). However, after omitting only one outlier (rs671, which has a strong relationship with alcohol drink), the significance for horizontal pleiotropy resolved. A distinct FBS / SBP trajectory (controlled and uncontrolled groups) over time was confirmed using latent group trajectory analysis after selecting the healthy population at baseline without T2D and hypertension. Subsequently, the incidents of hypertension / T2D were evaluated according to each FBS / SBP trajectory using Cox proportional hazard regression. There was no significant difference between the FBS uncontrolled and controlled groups after adjusting for covariates including antidiabetic medications. Conversely, a significantly higher risk of increased SBP was detected in the uncontrolled group relative to the controlled group (HR=1.27, 95%CI: 1.16-1.38 after adjusting for covariates including antihypertensive medications. Conclusion: A bidirectional causal association between fasting blood sugar level and systolic blood pressure in the Korean general population was identified based on the life course approach. In the future, elaborative large biobank studies including countless genetic variants and different environmental interactions are needed to validate the bidirectional causal association between FBS and SBP using the life course approach.