Assessing the linear and non-linear association of HbA1c with cardiovascular disease: a Mendelian randomisation study

Abstract

Aims/hypothesis: We aimed to evaluate whether genetically predicted HbA1c has an effect on the risk of cardiovascular diseases and investigate the shape of the relationship of genetically predicted HbA1c with cardiovascular diseases. Methods: We performed linear univariable, multivariable and non-linear Mendelian randomisation analyses in 373,571 white British participants (mean age 56.9) from the UK Biobank. Results: In univariable linear Mendelian randomisation analysis, a 1 mmol/mol increase in genetically predicted HbA1c was associated with higher risk of coronary artery disease (OR 1.03, 95% CI 1.02, 1.05), stroke (OR 1.02, 95% CI 1.00, 1.05) and hypertension (OR 1.02, 95% CI 1.01, 1.03). Multivariable Mendelian randomisation adjusted for the effect of haemoglobin gave a consistent conclusion for coronary artery disease. The associations with stroke and hypertension were directionally similar but with wider CI overlapping the null. Non-linear Mendelian randomisation indicated that the shape of the effect of genetically predicted HbA1c on cardiovascular outcomes was likely linear. Conclusions/interpretation: The study suggests a detrimental effect of HbA1c on coronary artery disease in both men and women, and the effect is via a glycaemic characteristic. The shape of the genetic association of HbA1c with these cardiovascular outcomes, in particular coronary artery disease, is likely to be linear. Graphical abstract: [Figure not available: see fulltext.]