Medical Therapies for Prevention of Cardiovascular and Renal Events in Patients With Atrial Fibrillation and Diabetes Mellitus

Laurent Fauchier; Giuseppe Boriani; Joris R. deGroot; Reinhold Kreutz; Peter Rossing; A. John Camm


Europace. 2021;23(12):1873-1891. 

In This Article

Clinical Complications in Atrial Fibrillation With Diabetes

The Increased Risk of Cerebrovascular and Cardiovascular Complications When Diabetes and Atrial Fibrillation Coexist

Diabetes mellitus is a disease linked to lifestyle, and therefore usually associated with additional CV risk factors for stroke such as obesity, hypertension, and dyslipidaemia, thus creating a complex interplay of adverse influences resulting in an increased risk of cerebrovascular events.[48] Both T2DM and T1DM are associated with an increased risk of stroke, even if the risk may be lower in T1DM.[49] In epidemiological studies performed in different populations, the relative risk of stroke in DM ranged between 1.4 and 5.8,[49] being 2.5 for males and 3.6 for females in the 20-year follow-up of the cohort of the historical Framingham study, aged 30–62 years at study entry.[50]

The patterns of strokes associated with DM are different from strokes associated with AF since even if most of the strokes in diabetic patients are ischaemic, and not haemorrhagic, the most common type corresponds to lacunar infarcts (i.e. small 0.2–15 mm, non-cortical infarcts), related to microvascular disease and the co-existence of hypertension.[51] Moreover, the strokes associated with DM have an increased risk of mortality, as well as more stroke recurrences and stroke-related dementia as compared with non-diabetic patients.[49]

In view of the interplay between AF and DM, with around 15% of patients with DM presenting with AF (Figure 2), according to data from a meta-analysis on T2DM[9] and around 30% of AF patients affected by DM,[52,53] it is difficult to assess the effects of DM control and antidiabetic drugs on stroke risk in patients with AF. The role of glycaemic control and duration of DM on stroke risk has been object of several investigations. In an analysis based on the Danish National registry, Overvad et al.[54] found that duration of DM, when analysed as a continuous variable, was associated with the risk of stroke/thromboembolism in a dose–response-dependent manner but was not associated with a higher risk of bleeding during anticoagulant treatment. In an analysis of a group of patients enrolled in the Anticoagulation and Risk Factors in AF (ATRIA) study, focusing on periods where patients were not anticoagulated,[55] an increased rate of ischaemic stroke was found in a relationship with a longer duration of DM [adjusted hazard ratio (HR) 1.74; 95% CI 1.10–2.76] but no significant relationship with increased HbA1c. The latter finding is in contrast with what was reported by Saliba et al.[56] who found a clear dose–response relationship between HbA1c and risk of incident stroke/transient ischaemic attacks, in a relatively short follow-up period (<1 year) (Figure 3).

Figure 3.

Stroke risk assessment in patient with diabetes and no formal stroke risk factors CHA2DS2-VASc = 1 in men and 2 in women. AF, atrial fibrillation; BMI, body mass index; DM, diabetes mellitus.

The complexity of the relationship between glycaemic control and stroke risk is highlighted by an additional analysis of the Danish registry[57] where, in patients with incident AF and T2DM, increasing levels of HbA1c were associated with a higher risk of thromboembolism, but this association was not confirmed when DM duration was ≥10 years. Recent data from Taiwan[58] indicate that in diabetic patients with AF the risk of stroke/systemic embolism is significantly increased once HbA1c levels exceed the value of 6.5%, among patients not treated with oral anticoagulants according to guidelines. The risk of major bleeding was comparable across all HbA1c levels categories.

Independently to the extent of glycaemic control, the data from PREFER in AF (European Prevention of Thromboembolic Events–European Registry in AF) showed that the risk of stroke/thromboembolism was the highest in individuals with DM treated with insulin compared with those not treated with insulin (5.2% vs. 1.8%; HR 2.96; 95% CI 1.49–5.87) and compared with non-diabetic patients.[59] Whether this association reflects a causal effect or a clustering of risk factors and long DM duration in patients treated with insulin compared to other glucose-lowering agents is not clear. Apart from insulin, the effects of other drugs for glycaemic control on stroke risk is still object of investigation, with a lower risk of stroke when using metformin[60] and thiazolidinediones.[61]

Beyond stroke, DM is associated with worse outcomes in patients with AF enrolled in contemporary registries.[31] In interpreting these associations, the higher prevalence of persistent and permanent AF, as well as the higher prevalence of HF, CKD, and coronary artery disease (CAD), among patients with DM has to be considered. The worse profile of diabetic patients with AF was also confirmed in EORP-AF (EURObservational Research Programme Atrial Fibrillation) General Registry, where diabetic patients had a higher prevalence of co-morbidities and a higher occurrence of all-cause, CV, and non-CV mortality at 1 year.[62] The reciprocal relationship was also found in DM patients, in whom AF was associated with worse outcomes. In the ADVANCE study, diabetic patients with AF had increased risks of major coronary events, stroke, HF, CV death, and all-cause mortality compared with T2DM patients without AF.[6]

Diabetic retinopathy is a typical microvascular complication of DM and in the Loire Valley AF project[63] crude rates of stroke/thromboembolism increased in a stepwise fashion when patients without DM and with AF were compared with patients with DM with no retinopathy and patients with DM with retinopathy. However, the presence of diabetic retinopathy did not emerge as an independent predictor for stroke/thromboembolism or severe bleeding on multivariate analysis.

Diabetes Subtypes, Diabetes Severity, and the Risk of Atrial Fibrillation

Diabetes mellitus is a heterogeneous chronic metabolic disease that is characterized by hyperglycaemia which may be caused by several different mechanisms, such as low insulin production and/or decreased insulin sensitivity. T1DM (10–15% of cases) often has a young age of onset and is characterized by a lack of insulin production that is associated with autoimmunity, whereas T2DM (80–85% of cases) has an older age of onset and is characterized by abnormally high circulating insulin levels that are associated with obesity-related insulin resistance.[64] Furthermore, additional rare forms of DM exist, which often have a genetic background, such as maturity-onset DM of the young (MODY) and neonatal DM.[65]

The risk for AF has been found increased both for T1DM and TT2DM. The risk in T1DM was studied in a cohort of 36 258 patients with T1DM the Swedish National Diabetes Register, matched with 17 980 controls followed for approximately 10 years.[3] The excess risk of AF in individuals with T1DM increased with worsening glycaemic control and renal complications. Among individuals with normo-albuminuria, no excess risk of AF was noted in men with T1DM who had HbA1c lower than 9.7% (<83 mmol/mol) or in women with T1DM who had HbA1c lower than 8.8% (<73 mmol/mol).

In a similar study of the risk for AF in T2DM, a total of 421 855 patients with T2 DM from the Swedish National Diabetes Register and 2 131 223 controls from the Swedish Population Registry were included and followed for 13 years.[66] In the fully adjusted Cox regression, the risk of T2DM on incident AF was 28% greater vs. controls, HR 1.28; 95% CI 1.26–1.30, P < 0.0001. The excess risk of AF in individuals with T2DM increased with worsening glycaemic control and renal complications. For individuals with HbA1c ≤ 6.9% (≤ 52 mmol/mol) and normo-albuminuria the excess risk vs. controls was still increased, adjusted HR: 1.16; 95% CI, 1.14–1.19 (P < 0.0001).[66]

When applying the CHA2DS2-VASc score to assess risk for stroke in AF the D (diabetes) is simply defined as fasting blood glucose ≥7 mmol/L (or ≥126 mg/dL), or treatment with oral hypoglycaemic drugs and/or insulin.[41] It has been included in the 2020 ESC Guidelines for the diagnosis and management of AF that the excess risk for stroke associated with AF and DM, is very similar in T1DM and T2DM except perhaps for a slightly increased risk in T2DM compared to T1DM in patients <65 years of age.[41] This was based on a study using data from Danish nationwide registries to identify patients with a prior diagnosis of DM and an incident AF diagnosis from 2005 to 2015, who were followed with thromboembolism as the outcome. The study population included 10 058 people with a prior diagnosis of DM and an incident diagnosis of AF. At 3-year follow-up, there was no difference in the risk for thromboembolism in T2DM compared to T1DM, with an adjusted HR 1.15, 95% CI 0.91–1.44. In an age-stratified analysis, patients with T2DM aged below 65 years of age had an adjusted HR 1.97, 95% CI 1.07–3.61 compared to T1DM, whereas there were no differences in the older populations.[67]

In addition to the DM types, DM is heterogeneous with different degrees of target organ damage affecting eye, nerves, and kidney, which affects the associated risk for CV complications.[32,68] The risk for stroke in subjects with DM and AF increases with longer DM duration, as well as with more DM co-morbidities such as nephropathy and retinopathy.[63]

The impact of DM therapy on the risk for AF has been debated (Figure 4). It has been suggested that metformin and pioglitazone may reduce the risk for AF.[69] SGLT2 inhibitors compared to placebo were associated with more new-onset AF in EMPA-REG OUTCOME (empagliflozin vs. placebo), but fewer incident AF cases in CANVAS (canagliflozin vs placebo) and DECLARE TIMI-58 (dapagliflozin vs. placebo).[70–73]

Figure 4.

Risk of incident AF with use of glucose-lowering therapies in recent meta-analyses or large observational analyses. AF, atrial fibrillation; CI, confidence interval.

In the DECLARE TIMI-58 trial, dapagliflozin reduced the risk of AF/AFL events by 19% (264 vs. 325 events; 7.8 vs. 9.6 events per 1000 patient-years; HR 0.81, 95% CI 0.68–0.95, P = 0.009). The reduction in AF/AFL events was consistent regardless of the presence or absence of a history of AF/AFL or atherosclerotic CVD at baseline.[73] These findings were exploratory but are relevant with increasing use of SGLT2 inhibitors in the treatment of T2DM, particularly in the presence of HF or CKD.[74] Glucagon-like peptide-1 receptor antagonists (GLP1-RA) have been investigated in various populations with an increased CV risk, but only a few studies reported new onset of AF with no significant differences when using GLP1-RA.[75,76]

The sub-classification of DM types is more complex than distinguishing just two types. Recently, Ahlqvist et al.[77] presented a novel data-driven approach to subclassification. They identified five patient clusters in 8980 adults with newly diagnosed DM. The cluster analysis was based on six preselected parameters: age at diagnosis; body mass index (BMI); presence or absence of glutamate decarboxylase antibodies (GADA) to identify autoimmune DM; glycated haemoglobin A1c (HbA1c) levels to assess glycaemic control; homoeostatic model assessment 2 (HOMA2)-B to assess β-cell function on the basis of C-peptide and glucose concentration; and HOMA2-IR to assess insulin sensitivity. The subtypes has subsequently been confirmed in other studies.[78] In relation to CV complications, the subtypes seem to be important as the severe insulin-resistant diabetes (SIRD) cluster had the highest risk of coronary events (HR 1.76, 95% CI 1.36–2.27) and stroke (HR 1.72, 95% CI 1.20–2.48) compared with the mild age-related diabetes (MARD) cluster,[77] but the importance in relation to AF and stroke risk is not known. Finally, gestational DM defined as glucose intolerance with onset or first recognition during pregnancy, which occurs in approximately 3–4% of all pregnancies, have not as such been linked to AF, but the presence of gestational DM increases the risk for later development of T2DM significantly.[79]