The impact of glycated hemoglobin A1c on cardiovascular risk in patients with and without coronary artery disease

Abstract Background Diabetes is defined according to the level of glycated hemoglobin (HbA1c) based on the risk of microvascular complications. However, HbA1c has been associated with increased risk of macrovascular complications at lower HbA1c levels than the conventional cut-off for "microvascular" diabetes. In patients without known diabetes, it is unsettled when HbA1c becomes a marker of macrovascular complications in non-diabetes patients with and without coronary artery disease (CAD). Purpose We tested the hypothesis that the HbA1c cut-off for macrovascular complications differ for patients with versus without CAD. Methods We included every patient undergoing coronary angiography in Western Denmark from 2011-2021 without known diabetes prior to examination. Follow-up was censored if it exceeded 10 years. We examined the risk of atherosclerotic cardiovascular disease (ASCVD, a composite of myocardial infarction and ischemic stroke) and HbA1c using Cox regression. We modeled the relation between ASCVD and HbA1c using restricted cubic splines with 38 mmol/mol as reference. In a separate analysis, we stratified patients by HbA1c levels (<30 mmol/mol, 31-38 mmol/mol (reference), 39-47 mmol/mol, and ≥48 mmol/mol) and CAD. Results We included 56,927 patients without known diabetes at the time of angiography, of whom 1,225 (2.2%) had undiagnosed, untreated diabetes, i.e. HbA1c ≥48 mmol/mol. Mean HbA1c was 38 mmol/mol. In patients without CAD we did not observe any association between HbA1c and ASCVD (Figure 1). In contrast, we observed a U-shaped relationship between HbA1c and ASCVD in patients with CAD. The lowest risk of ASCVD was observed at HbA1c 35 mmol/mol, which was followed by a steady incline with higher HbA1c levels. When categorizing HbA1c, the adjusted ASCVD risk was significantly increased in patients with CAD and HbA1c 39-47 mmol/mol (aHR 1.09, 95% CI 1.01-1.19), with HbA1c ≥48 mmol/mol (aHR 1.60, 95% CI 1.32-1.94) having the highest ASCVD risk (Figure 2). Conclusions The current study provides two novel findings. Firstly, HbA1c provides no prognostic information concerning risk of ASCVD when CAD is absent. Secondly, a U-shaped relationship between HbA1c and ASCVD was found in patients where CAD is present. The lowest risk of ASCVD was observed at HbA1c 35 mmol/mol and the risk steadily increased with higher (and lower) HbA1c levels. Thus, the risk of ASCVD associated with HbA1c starts much earlier than the conventional 48 mmol/mol HbA1c cut-off for diabetes, but the relationship exists only when CAD is present.Figure 1Figure 2