What's new?

This study evaluates the association between the use of sodium‑glucose cotransporter 2 inhibitors (SGLT2is) and bleeding and other clinical outcomes in anticoagulated patients with atrial fibrillation (AF), with and without diabetes. Prespecified subanalyses were conducted in patients with AF and diabetes, as well as in nondiabetic patients with AF and heart failure (HF). This real‑world study examines the impact of SGLT2is exclusively in nondiabetic AF patients with HF for whom data are only just emerging. The findings suggest that the pleiotropic effects of SGLT2is may extend to a reduced risk of bleeding and potential antiarrhythmic effects.

Introduction

After years of serving as antidiabetic drugs, sodium‑glucose cotransporter 2 inhibitors (SGLT2is) expanded into the cardiovascular domain following evidence from cardiovascular outcomes trials (CVOTs) and subsequent randomized controlled trials (RCTs) that demonstrated significant benefits in terms of mortality and heart failure (HF) hospitalization reduction.^1-7 SGLT2is now form an integral part of guideline‑directed medical therapy in HF.

Recent evidence suggests that SGLT2is may have a role in reducing the risk of atrial fibrillation (AF). In an analysis from the DECLARE‑TIMI58 trial (Multicenter Trial to Evaluate the Effect of Dapagliflozin on the Incidence of Cardiovascular Events), treatment with dapagliflozin was seen to reduce the risk of incident AF/atrial flutter (AFl), as well as subsequent episodes of AF/AFl in those with and without a history of AF/AFl, even after considering factors such as atherosclerotic cardiovascular disease, history of HF, sex, and body mass index (BMI).⁸ In contrast, a substudy from the CREDENCE trial (Evaluation of the Effects of Canagliflozin on Renal and Cardiovascular Outcomes in Participants With Diabetic Nephropathy) showed no significant difference in AF/AFl risk with canagliflozin but the total number of events recorded in the trial was small (n = 115). When these results were pooled with those from the 3 other CVOTs, the reduction in the AF/AFl risk was found to be significant.⁹ In line with this, several meta‑analyses of RCTs and cohort studies have demonstrated a lower risk of incident AF associated with SGLT2is.^10-18 Though the aforementioned trials were not designed to assess AF as a prespecified end point, these data suggest that SGLT2is may have some relevance in modulating arrhythmia risk.

Further to the above, there is limited evidence to suggest that SGLT2is may be beneficial in patients who already have AF. In an analysis from the EMPA‑REG OUTCOME trial (BI10773 [Empagliflozin] Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients), empagliflozin had a more pronounced absolute treatment effect in patients with than without AF, indicating that they may stand to achieve more benefit.¹⁹ A holistic approach to the management of patients with AF has been put forward for several years, and SGLT2is may be relevant in this context considering the several possible benefits mentioned above.^20,21 However, the number of patients with pre‑existing AF in the available RCTs has been relatively small and at present, there is little information on how SGLT2is affect certain AF‑related adverse outcomes in this specific patient population. Thus, we sought to evaluate this in a real‑world population of AF patients, including individuals with and without diabetes.

Patients and methods

Study design and population

This retrospective, observational cohort study was conducted using the TriNetX research network, a global federated administrative database with real‑time updates of electronic medical records (EMRs). It holds data on approximately 85 million patients from over 70 health care organizations (HCOs) across 7 countries, with the majority of centers based in the United States. Other participating countries include Germany, United Kingdom, Italy, Singapore, and Israel. A detailed description of the database is provided by Topaloglu et al²² and can be found online (https://trinetx.com/company‑overview/).

Briefly, the TriNetX research network database encompasses anonymized EMRs of patients registered with the network, and holds information on patient demographics, clinical details including diagnoses, medications, and investigations as well as any procedures, from settings such as general practice surgeries, community and secondary hospitals, providing detailed real‑world data.

All anticoagulated patients with AF diagnosed between January 1, 2014 and December 31, 2020 were identified from the TriNetX database using the International Classification of Diseases, Tenth Revision (ICD‑10) code I48 for AF. The patients were categorized into 2 groups of individuals treated and not treated with SGLT2is (Figure 1). The searches were run on January 23, 2026. At the time of the search, there were 97 participating HCOs within the TriNetX research network, and 88 responded with the patient data.

As TriNetX only provides access to deidentified data, research studies conducted using the network do not require ethical approval. Nonetheless, extensive data quality assessments are performed to ensure adherence of the platform to the institutional review board requirements and conformance, completeness, and plausibility of the available data.^22,23 Study proceedings were carried out according to the Declaration of Helsinki.

Results

A total of 876 007 patients were initially identified. Of these, 789 758 were not on an SGLT2i (mean [SD] age, 67.8 [10.5] y) and 86 249 (mean age, 70.6 [10.4] y) were on an SGLT2i (Figure 1). The patients in the SGLT2i group were older, less often women, and were more likely to have cardiac risk factors, such as hypertension, ischemic heart disease and HF, as well as comorbidities such as diabetes, liver and kidney disease than those not treated with SGLT2is. This group also had a higher proportion of patients with a history of bleeding and predisposition to falls, was more likely to have undergone cardiac procedures, be prescribed other cardiac and antidiabetic medications, and had poorer blood sugar control and triglyceride levels and higher BMI in comparison with the non‑SGLT2i cohort (Table 1).

After PSM, each group had 51 320 patients (mean [SD] age, 70.1 [9.1] y in the non‑SGLT2i group and 70.2 [10.6] y in the SGLT2i group). The outcomes were evaluated in the matched cohorts and are summarized in Table 1 with HRs for risk illustrated in Figure 2.

Any bleeding

There were 3848 and 6057 bleeding events observed in the SGLT2i and non‑SGLT2i groups, respectively. The SGLT2i use was associated with a lower risk of bleeding (HR, 0.669; 95% CI, 0.642–0.697; log‑rank P <⁠0.001). The Kaplan–Meier curve for bleeding is shown in Figure 3A.

Discussion

In this real‑world study, the SGLT2i use was associated with a significantly lower risk of AF‑related complications, such as bleeding, hospitalizations for AF/AFl, need for cardioversion / ablation, and VAs and cardiac arrests. Second, reduced risks of all‑cause mortality, IS/TIA, hemorrhagic stroke, incident HF, MI, and TEs were also observed with the SGLT2i use, as compared with their nonuse in this population of anticoagulated AF patients. Indeed, these observations suggest that SGLT2is might be useful in the holistic approach put forward over the last decade in the guidelines for the management of patients with AF.^20,21

To our knowledge, the effect of SGLT2is on the bleeding risk in AF has not been specifically evaluated in a large, observational study. In our analysis, the SGLT2i use was associated with a 33% reduction in the risk of any bleeding. The bleeding risk, aside from a stroke risk, is the single most important factor that influences decisions around anticoagulation, and factors such as uncontrolled hypertension and renal disease can significantly elevate the likelihood of a bleed.

SGLT2is have been shown to have a blood pressure lowering effect by improving endothelial function and arterial stiffness and through inactivation of the renin‑angiotensin‑aldosterone system, resulting in arterial vasodilation and reduced afterload.²⁴ Several meta‑analyses have demonstrated significant reductions in blood pressure with SGLT2is.^25,26 In 1 study, the greatest reductions in pulse pressure and mean arterial pressure were observed in the oldest patients with the highest baseline systolic pressure (ie, those who are most prone to bleeding should they receive anticoagulation).²⁷ SGLT2is may also reduce the predisposition to bleeding associated with anticoagulation and aid hemostasis by lowering the risk of acute kidney injury, including that requiring dialysis, and slowing progression of chronic kidney disease.^28,29

AF is closely linked with endothelial dysfunction, a key component of the Virchow triad for thrombogenesis associated with several adverse outcomes, such as thromboembolism, abnormal hemostasis, and bleeding.³⁰ SGLT2is have been shown to decrease the levels of reactive oxygen species and proinflammatory cytokines, which can reverse endothelial dysfunction, consequently inhibiting vascular inflammation and platelet activation, and decreasing the risk of not just bleeding but also of atherosclerotic and thrombogenic events.^31,32 This may explain the reduced risk of ischemic and hemorrhagic strokes, MI, and TEs observed in our study. SGLT2is were associated with a 20%–25% lower risk of IS/TIA in the entire AF as well as AF‑diabetes and AF‑HF cohorts. Similar decreases have been reported in the few studies that have examined the association between this outcome with the SGLT2i use in patients with pre‑existing AF,^33-35 except for a single study where no significant difference was observed.³⁶ In addition, our findings showed a 31% reduction in the risk of hemorrhagic stroke, consistent with findings from existing meta‑analyses (though not exclusively based on AF patients).^37,38

Another key finding from our analysis was the reduction in AF/AFl hospitalizations and composite of cardioversions and ablations even after taking antiarrhythmic drugs into account. These findings are supported by several clinical studies^36,39-44 demonstrating a significantly lower risk of AF‑related health care utilization in terms of hospital visits, need for subsequent cardioversion, redo ablation, and new antiarrhythmic drug therapy post–catheter ablation as well as a reduced risk of AF recurrence with SGLT2i use, suggesting they may have antiarrhythmic properties. This is hypothesized by Paasche et al,⁴⁵ whose study showed that treatment with dapagliflozin was associated with a significant reduction in the excitability of cardiomyocytes and reduction in the myocardial conduction velocity, more pronounced in the atrial than ventricular cells.

In addition to the direct antiarrhythmogenic effects, SGLT2i‑induced decreases in the level of epicardial fat and proinflammatory adipokines, which contribute to cardiac remodeling, processes involving intracellular sodium / calcium homeostasis that may increase threshold for arrhythmic triggers and off‑target actions, such as reductions in plasma volume and glucose‑induced inflammation, may also play a role.^46,47 It is unlikely that arrhythmia risk mitigation solely occurs due to processes that take place in isolation within the cardiovascular system. Disease modification and prevention resulting from SGLT2i impacting various pathways within the renal, endocrine, and metabolic systems are likely to mediate this risk. Of course, it is entirely possible that SGLT2is have no bearing on the arrhythmia risk but offer better symptom control in AF patients, reducing their need for further treatment and intervention. It is well evidenced that AF management addressing symptom control is associated with better outcomes.^48,49

Given the lack of SGLT2 proteins in the heart, it is believed that the direct antiarrhythmic effects of SGLT2is are executed through their action on SGLT1 proteins found in cardiomyocytes.^47,50 Drugs such as dapagliflozin and empagliflozin have been shown to have high SGLT2/SGLT1 selectivity, whereas canagliflozin is comparatively less selective. This would imply that canagliflozin has a greater antiarrhythmic potential than empagliflozin and dapagliflozin but the latter appears to have the most evidence as an antiarrhythmic, with some studies even suggesting that the reduced risk of arrhythmias is driven by dapagliflozin alone.⁵¹ Currently, it is unclear if certain drugs within this class are more likely to exert these effects.

There is inconsistent evidence regarding the risk of VAs and sudden cardiac death (SCD) associated with the SGLT2i use. In a meta‑analysis by Sfairopoulos et al,⁵² including 19 RCTs and over 55 000 patients, no significant differences were observed in the risk of VAs or SCD; however, event numbers were small, with 174 in the SGLT2i group and 191 in the placebo group. The meta‑analysis by Liao et al⁵³ showed similar findings but a borderline lower risk of SCD was observed in patients with HF only, suggesting these drugs may be more advantageous in certain subgroups of patients.

In our study, SGLT2is provided a modest but significant reduction in the risk of VAs and cardiac arrests in the overall and AF‑diabetes cohorts but not the nondiabetic AF‑HF group. The numbers of ventricular arrhythmias and cardiac arrests recorded in our study were much larger, though it must be noted that cardiac arrests would have comprised cases of SCD and sudden death due to other etiologies.

At present, evidence supporting the use of SGLT2is in patients with AF primarily stem from retrospective studies on diabetic patients. Consequently, determining whether these observations are directly due to the effects of SGLT2is or to mediation of risk factors such as diabetes and associated complications, or even HF, which shares a bidirectional relationship with AF, is not possible. Our subanalysis of the AF‑HF cohort indicates that the SGLT2i use is associated with a significantly reduced risk of adverse outcomes, regardless of diabetes, and further studies like this and RCTs will inform whether or not SGLT2is can be integrated into AF management as a treatment pillar. Several RCTs are currently underway investigating the effects of SGLT2is on AF recurrence,⁵⁴ disease burden, and quality of life and atrial remodeling.⁵⁵ Mechanistic studies examining the underlying processes are also warranted for further insights.