Mendelian randomization analysis supports the causal role of dysglycaemia and diabetes in the risk of coronary artery disease

Ross S, Gerstein HC, Eikelboom J et al.,
Eur Heart J. First published online: 30 March 2015 DOI: http://dx.doi.org/10.1093/eurheartj/ehv083

Background

Prospective observational studies have revealed progressive associations between various measures of glycaemia and cardiovascular (CV) outcomes, in people with or without a history of diabetes or CV events [1,2]. Randomised controlled trials evaluating the effect of glucose lowering have, however yielded mixed results, thus the etiologic relationship between dysglycaemia and coronary artery disease (CAD) remains unclear.
Mendelian randomisation analyses uses genetic associations to explore the effects of modifiable exposures on outcomes, by benefitting from the principle that genetic variants are randomly allocated at birth [3]. Mendelian randomisation analysis may help clarify the relationship between glucose traits, diabetes and risk of CAD. Genetic variants associated with dysglycaemia may, however, also be correlated with other CAD risk factors and reverse causation may be at stake. A method has been proposed by Do et al. to adjust Mendelian randomisation analyses to genetic effects on other risk factors, thereby allowing for the dissection of causal influence of different risk factors for the risk of CAD [4].
This study identified genetic variants associated with fasting glucose (FG), HbA1c and diabetes, and analysed whether the genetic effect of these dysglycaemia-related indices supports a causal association with CAD. In genome-wide databases, thirty single nucleotide polymorphism (SNPs) were found to be associated with FG, nine with HbA1c and 59 with diabetes.

Main results

• Linear regression analysis showed that SNPs associated with HbA1c (OR: 1.53 per % increase in HbA1c, 95%CI: 1.14-2.05, P=0.023) and diabetes (OR: 1.57, 95%CI: 1.16-2.05, P=0.008) predict risk of CAD, while SNPs associated with FG did not.
• After adjustment for the effects of SNPs on other CAD risk factors, only SNPs associated with diabetes remained significantly associated with CAD (OR: 1.63, 95%CI: 1.23-2.07, P=0.002). SNPs for FG and HbA1c did not show significant associations (FG: OR: 1.16, 95%CI: 0.94-1.42, P=0.185, HbA1c: OR: 1.66, 95%CI: 0.44-6.35, P=0.510).
• The Emerging Risk Factor Collaboration (ERFC) previously reported a significant correlation between diabetes and CAD (HR: 2.00, 95%CI: 1.83-2.19), and trends for FG (HR: 1.02 per mmol/L, 95%CI: 1.02-1.03) and HbA1c (HR: 1.43, 95%CI: 1.07-1.91) in individuals without diabetes or CAD in 73 prospective studies. The causal effect of diabetes on CAD risk revealed by Mendelian randomisation did not differ significantly from the risk estimate of ERFC, nor did the effects of FG and HbA1c (all P for difference > 0.05).
• An analysis explored whether sets of genes known to influence either β-cell function or insulin resistance showed different associations with CAD. SNPs influencing β-cell function (OR: 1.83, 95%CI: 1.19-2.62, P=0.015) and insulin resistance (OR: 2.35, 95%CI: 1.46-3.53, P=0.01) were both associated with an increased risk of CAD.

Conclusion

This Mendelian randomisation analysis supports a causal role of diabetes for CAD, based on using genetic information from 59 SNPs with known association with diabetes. These results are consistent with findings from previous observational studies. Restricting the analysis to genes affecting either β-cell function or insulin resistance, confirmed the role of diabetes and suggests that therapeutic interventions targeting either one of these different pathways may reduce CAD. Improved glycaemic control among diabetes patients and prevention of diabetes may lower CAD risk.

Editorial comment [5]

“Ross et al. have now published what will become a classic paper in the field of diabetes and cardiovascular risk. Using the powerful technique of Mendelian randomization, these authors show that a lifetime of lower glucose levels is indeed associated with a lower risk of adverse cardiovascular outcomes.” (…)
“The authors carefully adjusted for genetic effects of the diabetes associated SNPs under study on other cardiovascular risk factors such as hypercholesterolaemia, hypertension, and obesity”. (…) “Not all potential confounders were available for adjustment. For example, smoking and waist-to-hip ratio were not incorporated”. (…) “Also, it is worth mentioning that the patients included
were largely of European ancestry, so similar analyses in other racial and ethnic groups would be of added value.” (…)
“It is noteworthy that in the current study the degree of risk associated with diabetes is similar to what has been noted in prior observational analyses and meta-analyses, though with different populations and statistical techniques.” (…) “Interestingly, in the analysis by Ross et al., there were no significant differences according to whether the effect of the genetic loci was on beta-cell function or insulin resistance. This finding provides strong indirect evidence that the exact mechanism of glucose lowering may not matter, again assuming it can be accomplished safely.”

Find this article online at European Heart Journal

References

1. Sarwar N, Gao P, Seshasai SR et al. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet 2010;375:2215–2222.
2. Di AE, Gao P, Khan H et al. Glycated hemoglobin measurement and prediction of cardiovascular disease. JAMA 2014;311:1225–1233.
3. Smith GD, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003;32: 1–22.
4. DoR, Willer CJ, Schmidt EM et al. Common variants associated with plasma triglycerides and risk for coronary artery disease. Nat Genet 2013;45: 1345–1352.
5. Bhatt DL. Yes, hyperglycaemia is indeed a modifiable cardiac risk factor: so says Mendel. European Heart Journal. March 31 2015. doi:10.1093/eurheartj/ehv094