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

Anticoagulation and its Effects on Morbidity and Mortality (in Diabetes With Atrial Fibrillation)

Treatment with an oral anticoagulant of patients with AF at high risk of stroke leads to a substantial reduction in stroke and systemic embolism (SE).[80] Original data concerned vitamin K antagonists (VKAs), particularly warfarin. It is nowadays universally agreed that the efficacy of anticoagulant therapy is far superior than an antiplatelet agent and aspirin should not be used for the purpose of reducing the risk of AF-associated stroke/SE.[41] Oral anticoagulation is the main part in the 'A' pillar of the 'ABC' AF management ('A': Anticoagulation/Avoid stroke).[41]

Patients with AF can be stratified according to their risk for stroke and systemic embolism. The preferred cardioembolic risk stratification scheme (recommended by the ESC and ACC/AHA guidelines) is now the CHA2DS2-VASc system, particularly to exclude from anticoagulation those at sufficiently low risk of stroke.[41,81] This scoring system includes DM as a risk factor for stroke because the likelihood of developing a thrombo-embolic event is substantially increased in diabetic patients with AF. The use of VKAs in patients with DM is effective in reducing stroke and systemic embolism but anticoagulation control is more difficult to achieve than in non-diabetics[82,83] and adverse bleeding events are more likely in diabetics.

The development of DOACs included evaluation in five pre-approval Phase III randomized controlled trials (RCTs). In four of these, the DOAC was compared with dose-adjusted warfarin: dabigatran in the RE-LY (Randomized Evaluation of Long Term Anticoagulant Therapy) trial, rivaroxaban in the ROCKET-AF (Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation) trial, apixaban in the ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation) trial and edoxaban in the ENGAGE-AF (Effective aNticoaGulation with factor XA next GEneration in Atrial Fibrillation-Thrombolysis In Myocardial Infarction 48) trial. In the AVERROES (Apixaban vs. Acetylsalicylic Acid (ASA) to Prevent Stroke in Atrial Fibrillation Patients Who Have Failed or Are Unsuitable for Vitamin K Antagonist Treatment) trial, apixaban was compared with aspirin in patients who were unable to use warfarin. In each of the trials comparing a DOAC with warfarin, the results have been reported for the trial as a whole and in a subgroup of diabetic patients (Figure 5).

Figure 5.

Hazard ratios with DOACs compared to warfarin in patients with atrial fibrillation according to the presence of diabetes (blue symbols) or not (red symbols) in four randomized controlled trials. CV, cardiovascular; DOAC, direct oral anticoagulants; ICH, intracranial haemorrhage; SE, systemic embolism.

In the RE-LY trial,[84] there were 4221 diabetic patients out of the total of 18 113 patients.[85] Co-morbidities such as coronary disease and peripheral vascular disease were more common in diabetic patients and the rates of stroke, systemic embolism, mortality, and major bleeding were higher. Those treated with dabigatran had a similar relative risk reduction of stroke/SE but absolute risk reductions were greater in diabetic than non-diabetic patients (dabigatran 110 mg: 0.59% per year vs. 0.05% per year; dabigatran 150 mg: 0.89% per year vs. 0.51% per year).

ROCKET-AF compared rivaroxaban against dose-adjusted warfarin in 14 264 patients in whom the mean CHADS2 score was 3.6.[86] Forty percent of the patients in ROCKET-AF were diabetic patients who were younger and more obese than their non-diabetic counterparts.[87] Stroke, mortality, and major bleeding were more common in diabetic patients but the primary efficacy endpoint (reduction of stroke/SE) and safety endpoints (major haemorrhage) were similar in diabetic and non-diabetic subjects.

A total of 18 201 AF patients with a mean CHADS2 score of 2.1 were recruited to the ARISTOTLE trial. There were 4547 (24.9%) patients with DM and a higher average CHADS2 score (2.9) in ARISTOTLE.[88] Compared with the whole trial population, diabetic patients treated with apixaban had significantly lower rates of death (HR 0.83, 95% CI 0.67–1.02), stroke/SE (HR 0.75, 95% CI 0.53–1.05), and major haemorrhage (HR 0.49, 95% CI 0.25–0.95).

ENGAGE AF TIMI-48 enrolled 21 105 patients at a moderate-to-high risk of stroke (CHADS2 2.8).[89] In the diabetic subgroup analysis of 7624 diabetic and 13 481 non-diabetic subjects, the diabetic patients were younger, more obese and had a higher CHA2DS2-VASc score than the non-diabetic. There was no difference in efficacy between diabetic and non-diabetic patients. However, the risk of major bleeding was greatest in insulin-dependent diabetic patients. Similarly, there was no statistically significant interaction concerning major safety endpoints. However, since the risk of stroke/SE and major bleeding were higher in the diabetic patients, the absolute risk reduction associated with edoxaban treatment was greater in diabetic compared with non-diabetic patients. Cardiovascular deaths and major bleeding were reduced in both diabetic and non -diabetic patients to a similar extent in the higher dose subgroups (P > 0.0.05).[90]

The AVERROES trial was terminated early, after the recruitment of 5599 patients because of the obvious efficacy of apixaban when compared with aspirin.[91] The patients in the trial included 1098 (19.5%) with DM but there has been no analysis of this subgroup except that DM remained a significant risk factor for stroke in those who were treated with apixaban (HR 2.15, 95% CI 1.00–4.35; P = 0.05).[92]

The results across all four DOAC vs. dose-adjusted warfarin trials have been the subject of several meta-analyses. Ruff and colleagues provided an analysis of these trials consisting of 42 411 participants who received a DOAC and 29 272 who received warfarin. When considering patients who received the higher dose in the four DOACs (RE-LY, dabigatran 150 mg b.i.d; ENGAGE-AF–TIMI 48, edoxaban 60 mg o.d.; ROCKET-AF, rivaroxaban 20 mg o.d. and ARISTOTLE, apixaban 5 mg b.i.d.) stroke/SE was reduced by 19% (HR 0.81, 95% CI 0.73–0.91; P < 0.0001) in those who received a DOAC and major bleeding and mortality were also reduced by 14%, P = 0.06 and 10% respectively (P = 0.0003). In this group, there were 40 454 non-diabetics and 18 086 with DM and the risk reduction for stroke/SE seen with DOAC treatment was similar for both groups (P-int = 0.73) and for major bleeding (P-int = 0.12).[93]

The diabetic subgroups of these trials have also been accumulated in a metanalysis.[94] In the higher dose groups (n = 58 634), there were similar relative risk reductions for both stroke/SE (DM HR 0.80, 95% CI 0.69–0.93; non-DM HR 0.80, 95% CI 0.69–0.93, P-int = 0.81) and major haemorrhage (DM HR 0.95, 95% CI 0.75–1.20; non-DM HR 0.83, 95% CI 0.55–1.24; P-int 0.37) associated with DOAC treatment. When individual DOACs were considered, there was no significant difference for stroke/SE reduction, but for major bleeding with apixaban, there was a trend towards a more pronounced reduction of bleeding in non-diabetic patients in ARISTOTLE (P-int ≤0.1).[95] In the four pivotal DOAC trials, Patti et al.[96] observed that vascular death rates were reduced in diabetic patients with DOAC treatment compared with warfarin (4.97% vs. 5.99% with warfarin; RR 0.83, 95% CI 0.72–0.96; P = 0.01), but not in non-DM, probably because the absolute death rate reduction was higher in the diabetic than in non-diabetic patients (1.02% vs. 0.27%).

In a further meta-analysis, Plitt et al.[97] also showed that in comparison to warfarin DOACs reduced both CV death and intracranial haemorrhage similarly in both diabetic and non-diabetic subjects. This analysis also emphasized that dabigatran was the only DOAC to significantly reduce stroke/SE in diabetic and non-diabetic subjects alike and edoxaban was the only DOAC to significantly reduce major bleeding in patients both with and without DM. From their analysis, they also pointed out that although diabetic patients treated with insulin had higher rates of bleedings and stroke/SE both rivaroxaban and edoxaban were effective in reducing these outcome events in insulin-dependent and non-dependent patients, but data for dabigatran and apixaban were lacking.

In addition to data from RCTs, several reports have appeared based on large registries. Any comparisons made from registry data require adjustments to balance confounding factors between groups that are compared but residual confounders may also be present that will potentially confuse the data as will many other biases in the data. Nevertheless, information from these so-called 'real-world' settings are valuable to assess risks and outcomes in non-clinical trial situations and to confirm or question differences between treatments or interventions previously studied in more formal trial environments. Registry data have generally confirmed the results from RCTs and the results of some selected studies are summarized below.

Hsu et al.[98] used propensity weighting to compare diabetic patients taking DOACS rivaroxaban (n = 320) or dabigatran (n = 322) with patients taking warfarin (n = 1899) using data from the Taiwan National Health Insurance Research Database. The results were very similar to the relevant RCT. Rivaroxaban was similar to warfarin. Compared with warfarin, dabigatran decreased the risk of all-cause mortality (HR 0.348, 95% CI 0.157–0.771) and gastrointestinal bleeding (HR 0.558, 95% CI 0.327–0.955). Interestingly, Korgaonkar et al.,[99] who studied elderly patients with AF and DM from the Medicare database—2291 propensity-matched pairs of who were taking warfarin or one of the four approved DOACs also confirmed the interesting observation that DOAC use was not associated with more major GI bleeding when compared to warfarin.

The largest observational study of anticoagulation in diabetic patients (n = 154 324) was recently reported by Lip et al.[100] who compared warfarin with three DOACs. Lower risk of stroke was seen with apixaban (HR 0.67, 95% CI 0.57–0.77) and rivaroxaban (HR 0.79, 95% CI 0.71–0.89) whereas dabigatran was associated with a similar risk of stroke (HR 0.84, 95% CI 0.67–1.07), both apixaban (HR 0.60, 95% CI 0.56–0.65) and dabigatran (HR 0.78, 95% CI 0.69–0.88) were associated with a lower risk of major bleeding but rivaroxaban (HR 1.02, 95% CI 0.94–1.10) was associated with a similar risk of bleeding when compared with warfarin. Cohorts were compared through propensity score matching, there were potential residual confounders and only statistical associations could be concluded, not causal relationships.

Baker et al. used MarketScan data to study 10 700 diabetic patients taking rivaroxaban and compared them to a matched group of 13 946 warfarin users. In addition to endorsing previous findings related to the reduction of MACE associated with DOAC use they also found that major adverse limb events were very reduced by treatment with rivaroxaban rather than with warfarin. The studied population was relatively high risk (CHA2DS2-VASc = 4) and 11% had peripheral artery disease. Rivaroxaban was associated with a 25% reduction in MACE and a 63% reduction of major limb adverse events, including an 80% reduction of major limb amputation (HR 0.20, 95% CI 0.06–0.69)[101] (Figure 6). The mechanism whereby rivaroxaban might reduce adverse limb events may be a reduction in AF-related systemic arterial embolization, allowing activation of matrix proteins that inhibit vascular calcification or counteracting atherothrombotic effects of DM.

Figure 6.

Effectiveness, safety, and major adverse limb events in large databases of AF patients with concomitant diabetes mellitus treated with direct oral anticoagulants or vitamin K antagonists. (A) Baker et al.; (B) Chen et al. DOAC, direct oral anticoagulants; ICH, intracranial haemorrhage; MACE, major adverse cardiovascular events; MALE, major adverse limb events; VKA, vitamin K antagonist.

Chan et al. extended this observation using data drawn from the Taiwan National Health Insurance Research Database. In this study of diabetic patients, 5812 taking warfarin were compared, using propensity weighting, with 20 967 patients taking a DOAC. Major adverse limb events (MALE) were 28% less in patients taking a DOAC (aHR 0.72, 95% CI 0.57–0.92; P = 0.0083) (Figure 6). Interestingly the advantage associated with DOAC treatment persisted in patients also taking antiplatelet drugs.[102] In this study, the favourable effect on limb events was not confined to rivaroxaban but seemed to apply to all four DOACs. Controlled trial data are needed to clarify the effect of DOACs in general and each DOAC in particular on limb amputation, etc. but none is yet available.

There is some limited vascular outcome data in the setting of a RCTs post-acute coronary syndrome/revascularisation. In the RE-DUAL (Randomized Evaluation of Dual Antithrombotic Therapy with Dabigatran vs. Triple Therapy with Warfarin in Patients with Nonvalvular Atrial Fibrillation Undergoing Percutaneous Coronary Intervention) trial setting, 2725 patients with AF of whom 993 had DM, dabigatran plus clopidogrel or ticagrelor, compared with conventional triple therapy, was equally effective at reducing bleeding and had comparable efficacy endpoints in diabetic and non-diabetic patients post angioplasty.[103] However, this trial was not designed to demonstrate effects specifically in diabetic patients and was underpowered for this purpose.

In a large meta-analysis, Abdool et al. added four observational studies to the results of randomized DOAC vs. warfarin trials to bring the total number of patients in their pooled analysis to 147 943. They confirmed previous observations that there was no interaction between DM status and adverse outcomes and added that absolute risk reduction was more in diabetic patients because of their higher untreated adverse outcome risk.[104]

Diabetic patients very often have some degree of renal impairment and frank CKD. Both DM and CKD share common aetiologies and DM is the single biggest cause of CKD. Diabetes mellitus increases the likelihood of stroke, major bleeding, and mortality in patients with AF, and CKD amplifies these risks. There is clear evidence from numerous studies that for mild and moderate CKD, DOAC therapy is at least non-inferior to and in many instances superior to dose-adjusted warfarin. Recently, a flurry of observational studies and at least one RCT (RE-LY) have raised the possibility that renal function may be progressively impaired when VKA therapy, as opposed to DOAC treatment, is prescribed.[105–107] Observational evidence also exists that AF patients with DM in addition to CKD can also be protected from progressive renal impairment by avoiding a VKAs and in favour of DOAC therapy (Figure 7).[108]

Figure 7.

Comparison of the progression of chronic kidney disease (CKD) to Stage 5 CKD/haemodialysis or development of acute kidney injury in atrial fibrillation patients treated with a direct oral anticoagulants (DOAC) or vitamin K antagonist (VKA). CI, confidence interval; HR, hazard ratio.
*Hernandez AV et al. Eur Heart J Qual Care Clin Outcomes 2019; pii: qcz047
**Vaitsiakhovich T et al. Presented at ESC 2019
***Bonnemeier H et al. Presented at ESOC 2019, Milan, italy, AS25–069

Since DM is part of most risk scoring schemes used to decide which patients with AF should be anticoagulated many diabetic patients do receive anticoagulation. However, because DM is probably the strongest of the individual risk factors allocated one point in the CHA2DS2-VASc score, one may advocate anticoagulating every diabetic patient with AF, especially when DM has been present for some time or is associated with an uncontrolled glycaemic status. Of course, those jurisdictions where the CHADS2 score is still in use, at least in part—for example, in Canada and Japan, most diabetic patients with AF are recommended for anticoagulation.

All the data suggest that when comparing dose-adjusted warfarin to DOAC therapy there is a similar relative risk reduction for all major efficacy outcomes including all-cause mortality, (cardio)vascular mortality, life-threatening bleeding, and intracranial haemorrhage in the diabetic and non-diabetic populations.[97,104] Because of the increased risk rate for these adverse events in diabetic patients a similar relative risk reduction generally translates into greater absolute risk reduction in the diabetic population.[104] As with non-diabetic patients, major bleeding and major gastrointestinal bleeding associated with different DOACs compared with warfarin seem to vary similarly in diabetic patients. For these reasons and because RCTs show better results with DOACs, DOAC drugs are preferable to VKAs for stroke prevention in DM as for the general population of at-risk AF patients.

There have been some observations suggesting differences between DOACs, such as the statistically significant reduction in vascular mortality associated with the use of rivaroxaban in diabetic patients and the far greater reduction of major bleeding seen with apixaban when used in non-diabetic rather than diabetic patients (Figure 5).[94,95,97] However, there is no head-to-head comparison between the different DOACs and even when pooling data from controlled trials and observational studies there is insufficient evidence to distinguish between DOACS concerning their suitability for the management of patients with AF and DM.

There is considerable speculation that DOAC treatment might suppress atrial fibrosis and collagen deposition induced by hypercoagulability.[109] Currently observational registries (e.g. RACE 5) and randomized clinical trial (e.g. ARTESiA [NCT01938248] and NOAH-AFNET 6 [NCT02618577]) are exploring whether AF will develop, recur or progress as quickly in patients treated with DOACS rather than VKAs. There is no conclusive data at present. Conversely, a recent registry study involving 10 746 patients with recent-onset AF reported that the development of diabetes was far less probable in the DOAC treated group than in those treated with VKAs (aHR 0.80, 95% CI 0.68–0.94; P = 0.007). This difference was seen with rivaroxaban, apixaban, and dabigatran, and was most obvious in patients older than 65 years and in those with good adherence to treatment.[110] This study could not demonstrate a causal relationship between treatment with a VKA and more development of diabetes but such an effect might relate to an increase of vitamin K-dependent osteocalcin which is known to stimulate insulin and adiponectin.

The recent revelation that diabetic AF patients treated with rivaroxaban, and possibly other DOACS, rather than warfarin are far less likely to suffer from major adverse limb events, including major amputation, adds to the conviction that DOAC treatment is far preferable to the use of VKAs in DM. Since CKD is a frequent comorbidity in patients with AF and DM and that treatment with a VKA leads to further progression of renal impairment, stroke prevention with DOAC therapy seems eminently preferable to the use of VKAs.