The relationship of blood glucose with cardiovascular disease is mediated over time by traditional risk factors in type 1 diabetes: the DCCT/EDIC study

Abstract

Aims/hypothesis: Chronic hyperglycaemia, as measured by HbA_1c levels, is a major risk factor for atherosclerosis and cardiovascular disease (CVD) in type 1 diabetes. Our aim was to describe the degree to which the effect of HbA_1c on the risk of CVD is mediated by its effect on traditional risk factors over time, and how these mediation pathways change over time.

Methods: The DCCT and its observational follow-up study, the Epidemiology of Diabetes Interventions and Complications (EDIC), followed 1441 participants for a mean of 27 years, with periodic measurement of HbA_1c and risk factors over time. We assessed the proportion of the HbA_1c effect on risk of CVD that was mediated through its effects on systolic BP (SBP), pulse rate, triacylglycerols and LDL-cholesterol (LDLc) levels, and how the proportion mediated changed over time.

Results: The association of HbA_1c with CVD outcomes was stable over time, while that of traditional risk factors (SBP, pulse rate, triacylglycerols and LDLc) increased. At 10 years of follow-up, the effect of HbA_1c on 10 year CVD risk was minimally mediated by SBP (2.7%), increasing to 26% at 20 years. Likewise, from 10 year follow-up to 20 year follow-up, the proportion of HbA_1c effect mediated through pulse rate increased from 6.3% to 29.3%, through triacylglycerols from 2.2% to 22.4%, and through LDLc from 9.2% to 30.7%.

Conclusions/interpretation: As participants age, the predictive association of mean HbA_1c on subsequent CVD events is increasingly mediated by its effect on standard risk factors. Thus, management of traditional non-glycaemic CVD risk factors may have increasing benefits in an ageing type 1 diabetes population with longstanding hyperglycaemia.

Trial registration: ClinicalTrials.gov NCT00360893 and NCT00360815.

Keywords: Area under the curve; Cardiovascular disease; Mediation proportion; Type 1 diabetes.

Fig. 1

The single mediator model, in which HbA_1c has a direct effect on CVD outcomes through the HbA_1c to CVD pathway, and an indirect effect on CVD is mediated by the effect of HbA_1c on SBP through the HbA_1c to SBP to CVD pathway. The total HbA_1c effect on CVD is the sum of the direct and indirect effects. The parameters α₁ and γ_1/γ₂ describe the effect of HbA_1c on SBP and the effects of HbA_1c and SBP on the risk of CVD, respectively (see ESM Mediation Analysis for details).

Fig. 2

AUC values for the updated mean SBP, LDLc levels, pulse rate or current triacylglycerol levels (Trigs), each adjusted for age and duration of diabetes (Model 1, diamonds) over (a–d) years 10–20 and (e–h) years 20–30 since enrolment. Also shown are the AUCs for models with age and diabetes duration alone (Model 2, circles), and age, diabetes duration and HbA_1c (Model 3, squares), and the full model (Model 4, triangles). The data presented for the models with age and diabetes duration (Model 2, circles), age, diabetes duration and HbA_1c (Model 3, squares), and the full model (Model 4, triangles) are identical in (a–d) and in (e–h)

Fig. 3

(a–d) Proportion of the effect of HbA_1c mediated over time through the updated mean (a) SBP, (b) LDLc, (c) triacylglycerols (Trigs) and (d) pulse rate in 10 year models (10–20, 11–21, etc) using covariate values at the start of each interval

Fig. 4

Decomposition of the total effect of HbA_1c on CVD risk into the direct and indirect effects mediated through multiple risk factors over time for the 10–20, 15–25 and 20–30 year models, using covariate values at 10, 15 and 20 years, respectively. Each shaded bar represents the proportion of the direct effect of HbA_1c on CVD risk (black) and the indirect effects through SBP (dark grey), pulse rate (light grey), triacylglycerols (white) and LDLc (diagonal stripes). Data are presented in Table 3