Typically, therapeutic anticoagulation is employed to achieve a reduction in risk of stroke. This risk is not trivial in the HF population.^3,4 Historically, the risk of stroke in the HF population was best explained by the elements of Virchow’s triad: blood flow abnormalities, vessel wall abnormalities, and abnormal blood constituents. While all the components of Virchow’s famous triad may apply to HF patients, it is the first component, blood flow abnormalities, that is presumed to play the biggest role in imparting stroke risk. Blood flow in HF is likely to be abnormal in the context of LV dysfunction (including regional areas of dyskinesis or aneurysm). Despite the plausibility of this physiologic explanation, investigation of the presumed hypercoagulable state in HF has not been confirmed in clinical trials.^5-10 The most recent, largest, and most rigorously designed of these was the Warfarin and aspirin in patients with heart failure and sinus rhythm (WARCEF) trial which was a randomized, double-blind, double-dummy study of 2305 patients worldwide.8 WARCEF demonstrated that in patients with LVEF ≤ 35% and no existing indication for anticoagulation (e.g., atrial fibrillation) there was no difference in the primary endpoint of ischemic stroke, intracerebral hemorrhage, or death from any cause between the groups randomized to aspirin versus those randomized to warfarin. Even when clinical trial data are combined, there does not appear to be a beneficial signal for the use of anticoagulation in HF without another risk factor (e.g., atrial fibrillation).^11-12 One possible explanation for this is an inherent increased bleeding risk in patients with HF, but this has not been confirmed empirically. The American College of Cardiology and American Heart Association (ACC/AHA) has therefore included a class IIb recommendation in the 2009 HF guidelines indicating that “the usefulness of anticoagulation is not well established in patients with HF who do not have atrial fibrillation or a previous thromboembolic event”.¹³ The Canadian Cardiovascular Society (CCS) and The European Society of Cardiology (ESC) make similar recommendations (Table 1).^14,15

Despite these findings and recommendations, a significant number of HF patients without another indication for anticoagulation continue to be prescribed therapeutic anticoagulation. Data spanning the last 20 years from registries and post-hoc analyses of clinical trials demonstrate that warfarin is prescribed to HF patients without another thromboembolic risk factor at a rate of between 8 and 17%.^10,16-¹⁸ However, importantly, none of the trials assessing anticoagulation in HF with sinus rhythm included direct-acting oral anticoagulants (DOACs). It remains to be seen if there is any benefit to anticoagulation with DOACs in HF and sinus rhythm, and if so, whether those benefits are balanced by acceptable bleeding risks. The COMMANDER HF study is an attempt to address this question of anticoagulation in HF with sinus rhythm in the age of DOACs.¹⁹ This ongoing study will assess the effectiveness and safety of rivaroxaban in reducing the risk of death, myocardial infarction or stroke in patients with HF and coronary artery disease.

Heart rhythm abnormalities – namely atrial fibrillation – are very common in the HF population with prevalence of atrial fibrillation estimated at 13-40%.^20,21 This prevalence of documented atrial fibrillation is considerable, but the true burden of atrial fibrillation in this population and others may be even greater owing to subclinical forms.²² The relationship between atrial fibrillation and HF is complex with both conditions acting as a risk factor and an outcome for the other. There is no doubt that HF is associated with increased risk of stroke in the presence of atrial fibrillation.^23,24 Indeed, assessment of stroke risk is conducted routinely by clinicians by using risk scores like the CHADS2 and CHA2DS2VASc, both of which include heart failure as a risk factor.

While atrial fibrillation clearly makes a large impact on stroke risk in HF, there are other stroke risk factors in HF that may be mitigated by therapeutic anticoagulation. Left ventricular dysfunction is one risk factor for the development of LV thrombus,^25-27 and acute myocardial infarction (MI) may increase risk of stroke in the post-MI period through multiple mechanisms including temporary or permanent reduction in LV function and LV thrombus development.²⁸ Indeed, experience has identified reduced EF following acute MI as one predictor of developing LV thrombus.²⁶ Experience from the 2002 Warfarin, Aspirin, or Both After Myocardial Infarction (WARIS II) trial demonstrated that anticoagulation following MI reduced the risk of the composite endpoint of death, nonfatal reinfarction or thromboembolic stroke.²⁹ Subsequent studies have shown an inconsistent effect of post-MI anticoagulation on the risk of death, stroke, or other important endpoints outcomes^26,30 Because these results are mixed, the guidelines are as well: the Heart Failure Society of America (HFSA) recommends anticoagulation in patients with a large anterior MI with ischemic cardiomyopathy while other groups do not (Table 1).³¹

Data supporting the use of therapeutic anticoagulation in patients with a documented LV thrombus are very limited in part because most data in this area were generated before the modern era of thrombolysis for acute myocardial infarction, percutaneous coronary intervention with dual antiplatelet therapy, stroke risk assessment with modern tools (e.g., CHA2DS2VASc), modern imaging techniques (e.g., cardiac MRI), and certainly before DOACs were available.²⁷ Nonetheless, various studies have demonstrated that the incidence of LV thrombus is still considerable compared to the precoronary intervention era.^26,32,33 In addition, for many providers, there is a lack of equipoise in treatment strategy when an LV thrombus is documented in a patient with HF, so a randomized trial of treatment strategies for documented LV thrombus may not be possible.

The need to balance risks of thromboembolism with risks of bleeding is paramount, but this balance is complicated by the fact that risk factors for stroke and for bleeding are frequently overlapping, especially in the HF population (e.g., advanced age). Moreover, evidence suggests that risk of stroke and hemorrhage while taking warfarin increase with increasing severity of heart failure. In an analysis of >62,000 HF patients taking warfarin for atrial fibrillation, the hazard for major bleeding in patients with the most severe HF versus least severe was 3.97 after adjusting for known risk factors; risk of stroke increased with increasing HF severity as well but less dramatically.³⁴

This conflict between bleeding and stroke risk is illustrated by common and validated risk scores used in clinical practice for estimating risk of stroke and bleeding, for example: CHADS2 and HAS-BLED, respectively.^23,35 Both risk scores include hypertension, stroke, and advanced age. (Other elements of the HAS-BLED score include abnormal renal/liver function, bleeding history or anemia, labile INR, and concomitant drug/alcohol use.) Thus, common baseline characteristics and comorbid conditions among HF patients simultaneously confer bleeding and stroke risk making clinical decisions surrounding anticoagulation complex. However, like the HAS-BLED risk score, other bleeding risk assessment tools have not specifically assessed the impact of heart failure on bleeding risk.^36-40 36-40 Further work is needed to evaluate how these bleeding risk assessment tools perform specifically in the HF population and in the era of DOACs.

The first DOAC to complete a phase III evaluation for nonvalvular atrial fibrillation was the direct thrombin inhibitor, ximelagatran.^44,45 However, after more than 7000 patients were randomized to warfarin vs ximelagatran, significant hepatotoxicity was noted, and this was a primary impediment to approval by the Food and Drug Administration (FDA). In 2010, dabigatran (Pradaxa®, Boerhinger Ingelheim), a second oral direct thrombin inhibitor, was the first DOAC approved by the FDA for stroke prophylaxis in nonvalvular atrial fibrillation. Approval of dabigatran was followed by the oral Factor Xa inhibitors: rivaroxaban (Xarelto®, Bayer pharmaceuticals) in 2011, apixaban (Eliquis®, Pfizer/Bristol-Meyers Squibb) in 2012, and edoxaban (Savaysa®, Daiichi Sankyo) in 2015. Approval of these drugs represents the experience of nearly 100,000 patients in published phase III randomized studies and many more in earlier phase investigation. The phase III experience included nearly 25,000 patients with heart failure as outlined in Table 2.^46-49

FDA approval of dabigatran was supported, in part, by the noninferiority RE-LY study of more than 18,000 patients randomized to warfarin versus two doses of dabigatran, 110 mg and 150 mg, respectively.⁵⁰ Patients with active liver disease or creatinine clearance less than 30 ml/min were excluded, and about one third of patients had HF. Ferreira and colleagues conducted a subgroup analysis to examine outcomes from RE-LY in patients with symptomatic HF.⁴⁶ The primary endpoints from the overall study were examined including: time to first occurrence of stroke or systemic embolism and time to first occurrence of major bleeding. Compared with warfarin, the hazard ratios for stroke or systemic embolism in the two dabigatran groups (110 mg twice daily and 150 mg twice daily, respectively) mirrored the results of the study overall. There was no statistically significant reduction in bleeding events in the HF subgroup taking either dose of dabigatran compared with warfarin. Importantly, there was no significant interaction between the treatment effect of either dose of dabigatran and the presence of HF in regard to the efficacy or safety endpoints. Of note, only 75 mg and 150 mg doses were approved by the FDA. Specific safety and efficacy data for the 75 mg dose are not available in the HF population but ongoing work within the FDA’s Mini-Sentinel initiative may shed light on this in the future.⁵¹

Patel and colleagues published ROCKET-AF in 2011 which was a multicenter, randomized double-blind, double-dummy trial at 1178 sites in 45 countries which examined safety and efficacy of warfarin versus rivaroxaban in nonvalvular atrial fibrillation.⁵² Patients were at high risk for thromboembolic events and important exclusions included those patients with significant liver disease and creatinine clearance less than 30 ml/min. A subgroup analysis of the HF population in the ROCKET-AF trial was performed by van Diepen and colleagues.⁴⁹ They found no statistically significant difference in the primary efficacy or safety outcomes between HF patients randomized to rivaroxaban versus warfarin. There was no interaction observed between the primary efficacy and safety endpoints and the presence of HF. In addition, when factors contributing to risk of stroke or systemic embolism within the HF subgroup were observed in isolation (LVEF, HF with preserved versus reduced systolic function, functional class or CHADS2 score), no interaction was seen suggesting that the benefit of rivaroxaban over warfarin extends to HF patients with AF over a broad range of risk.

In 2013, McMurray and colleagues reported on the HF subgroup from the ARISTOTLE trial48 which was a multicenter, double-blind double-dummy trial of patients randomized to apixaban or doseadjusted warfarin.⁵³ Like RE-LY and ROCKET-AF, patients with severe renal insufficiency (CrCl <25 ml/min, in this case) and active liver disease were excluded. They identified 2,736 patients from ARISTOTLE representing 19% of the total enrolled population who had an LVEF ≤ 40% or moderate or severe LV dysfunction. Patients with HF were quite different from those patients without HF who were older, less female, less ischemic, and less likely to have persistent or permanent AF as opposed to paroxysmal. HF patients in ARISTOTLE had a mean CHADS2 score of 2.22 compared to 1.88 for patients without HF. Patients with HF were more likely to experience stroke, systemic embolism, major bleeding, or death from any cause (HR 1.98 95% CI 1.77-2.22, p<0.0001). In keeping with the overall results of ARISTOTLE, apixaban was superior to warfarin for stroke or systemic embolism as well as bleeding outcomes, and there was no evidence of treatment heterogeneity according to presence of HF.

The primary phase III trial of edoxaban, ENGAGE AF-TIMI 48, included 8076 patients with HF defined as “current presence or history of clinical CHF class C or D”.⁴⁷ Patients with renal insufficiency (CrCl <30 m/min) were excluded from enrollment. Edoxaban became the newest member of the DOAC family with FDA approval in 2015. Interestingly, and unlike its DOAC predecessors, based on higher stroke or systemic embolism rates in patients with high normal or supranormal renal function, edoxaban is contraindicated in patients with CrCl >95 ml/min. Reports of safety and efficacy in the HF subgroup have been examined by Magnani and colleagues in a 2014 abstract.⁵⁴ They showed that there was no interaction between reduction in stroke or systemic embolism and the presence of HF.

A fourth oral factor Xa inhibitor, betrixaban, has been evaluated in a phase II trial for nonvalvular atrial fibrillation which demonstrated the drug to be well tolerated and safe.⁵⁵ More than a third of patients enrolled in Explore-Xa had a CHADS2 score of 3 or greater, but it is not known what proportion had HF. As yet, betrixaban remains unevaluated by the FDA, and there are no phase III trials registered with clinicaltrials.gov in the atrial fibrillation population.

As noted above, in all of the modern trials of anticoagulation with direct oral anticoagulants for stroke prophylaxis in atrial fibrillation, a significant proportion of patients with HF were enrolled. In the case of all of the FDA-approved DOACs specifically studied in a heart failure population (i.e., dabigatran, rivaroxaban, and apixaban) there has been no evidence of a significant interaction between safety and/ or efficacy outcomes and the presence of HF indicating that patients with HF should expect to benefit from the DOAC in a similar way as their counterparts without HF. Importantly, however, patients with significant renal and/or liver dysfunction were mostly excluded from DOAC trials, and these comorbidities are not uncommon among the heart failure population.^57-60 Indeed, in addition to the specific renal dysfunction groups excluded in each of the DOAC trials, the FDA labels indicate that edoxaban and rivaroxaban are contraindicated in patients with moderate to severe hepatic impairment, and apixaban is contraindicated in severe hepatic impairment. So, in a large portion of the HF population, warfarin remains the only effective option for therapeutic anticoagulation to reduce stroke risk.

Other comorbidities have been evaluated in the context of the DOACs as well. Extensive work by Lega et al examined the safety and efficacy of dabigatran, rivaroxaban, and apixaban in various comorbid subgroups.⁶¹ This group found that almost all subgroups had a treatment effect from the DOACs similar to the overall populations studied in the randomized clinical trials. It remains to be seen if these effects are borne out in clinical practice. In addition, the cardiovascular community awaits information on how HF and other important comorbid conditions impact the effects of edoxaban and betrixaban.

In addition to drugs to reduce stroke risk in HF, there are devicebased solutions that generally occlude or exclude the left atrial appendage (LAA). Two of the most widely known of these in the US are the Lariat® (SentreHEART, Redwood City, CA) and the Watchman® (Boston Scientific Corp, Marlborough, MA).^62,63 The Lariat enables the transcatheter ligation of the LAA via a combined transseptal and subxiphoid approach. Preliminary results which include a significant experience with HF patients demonstrate that procedural success can be high, but serious bleeding complications can occur; no long term safety and effectiveness data are available.⁶⁴ In addition, post procedural anticoagulation practices are widely variable,⁶⁵ and no unified recommendation exists. The Watchman device was recently FDA approved for occlusion of the LAA in patients with non-valvular atrial fibrillation based on two randomized non-inferiority trials in which 23 and 27% of subjects, respectively, had congestive heart failure.^62,63 In both cases the control groups were treated with adjusted-dose warfarin and noninferiority for stroke prevention was met, so the device is indicated for patients who are deemed warfarin candidates but have “an appropriate rationale” to seek a non-pharmacologic alternative.⁶⁶ It remains unclear how these devices and others will be incorporated with anticoagulants into routine clinical practice and whether safety and/or effectiveness will differ in the HF population.

Evidence supporting AF ablation in patients with HF is limited^67,73 as are the data for post-ablation anticoagulation in this population.^74,78 Currently, based on a paucity of high quality evidence, therapeutic anticoagulation with warfarin or a DOAC is recommended indefinitely in patients at high risk for stroke which would generally include those patients with HF.^79,80