Thrombo-embolism and antithrombotic therapy for heart failure in sinus rhythm. A Joint Consensus Document from the ESC Heart Failure Association and the ESC Working Group on Thrombosis

Epidemiological cohort data

A significant proportion of patients presenting with stroke or peripheral thrombo-embolism have HF, specifically HFrEF or asymptomatic echocardiographic evidence of LV systolic dysfunction (LVSD). About 14% of patients with stroke have HF,21 and ~20% of stroke patients have some evidence of LVSD (EF ≤50%).22 Nonetheless, data on thrombo-embolic risk in HF patients from large epidemiological studies to support this are limited, and many older studies do not differentiate between HFrEF and HFpEF. Also, many studies do not fully account for the risk of concomitant AF, and silent paroxysms of AF may occur frequently in HF patients.

What do the epidemiological studies tell us? In one analysis of 516 deaths in a long-term community study comparing HF patients with depressed LV function vs. those with preserved LV function, sudden death occurred in 21% vs. 16%, respectively.23 In patients with depressed LV function, new coronary occlusions [as reflected by myocardial infarction (MI)] occurred in only a small proportion of deaths according to poorly specified criteria during a 1.5-year follow-up.23

In the Rotterdam study of 7546 participants ≥55 years of age followed for 10 years, the risk of ischaemic stroke was increased more than five-fold in the first month after diagnosis of HF compared with no HF diagnosis, even after adjustment for age and sex [adjusted hazard ratio (HR) 5.79, 95% confidence interval (CI) 2.15–15.62], but this risk became attenuated over time (age- and sex-adjusted HR 3.50, 95% CI 1.96–6.25) after 1–6 months, becoming non-significant after the first 6 months following diagnosis (0.83, 95% CI 0.53–1.29, after 0.5–6 years).2 Adjustments for cardiovascular risk factors (including AF) modestly attenuated these HRs, but still showed a risk of ischaemic stroke comparable with that of controls only after 0.5–6 years (age- and sex-adjusted HR 0.58, 95% CI 0.37–0.92). Thus, the risk of ischaemic stroke appears strongly increased shortly after the diagnosis of HF, returning to normal within 6 months after onset of HF.2 HF patients also have a high rate of stroke recurrence and mortality after stroke;24,25 in a population-based study from Rochester, for example, this was 20% in the first year and 45% after 5 years.11 In the Framingham Heart Study,26 the risk of stroke was 4.1% per year for men and 2.8% per year for women with HF, although concurrent AF was present in many patients.

There is also a clear association between HF and VTE, with a two-fold increased risk, and important implications for prognosis and death.27,28

Case–control studies

In the Northern Manhattan Study (NOMAS), a multiethnic population-based case–control study of 270 patients with first ischaemic stroke and 288 matched controls, any degree of LV dysfunction was more frequent in stroke patients (24.1%) compared with controls (4.9%; P < 0.0001).22 Even a mild degree of LVSD (EF 41–50%) was still associated with an increased adjusted risk of ischaemic stroke.

Many publications from relatively small case–control or other observational studies from single centres and/or registries, with the inherent limitations of such studies, report wide variations in the incidence of arterial thrombo-embolism, ranging from 1.4% to 12.5%.293132333435363730–38(Table 1). Also, many of these studies included a mixture of dilated cardiomyopathy and ischaemic HF.

Further, the risk of stroke in HF patients is also directly related to the degree of LVSD, as seen in some retrospective studies (Table 1). Again, another limitation is that many of these studies included patients with AF, and thus the increase in the incidence of thrombo-embolic events could be more related to the inherent embolic risk in AF than in HF per se; conversely, anticoagulants given for AF with HF (and this was apparent in varying proportions, as high as 57%34) could result in a considerably lower incidence of thrombo-embolic events than when given for HF without AF. Most of the studies did not specify thrombo-embolism as an endpoint, and therefore the true incidence may be under-reported. Lastly, many studies do not differentiate between HFrEF and HFpEF.

Post-hoc analyses of trial data

Although in a selective cohort with unclear definition of MI, autopsy findings from the Assessment of Treatment with Lisinopril and Survival (ATLAS) study adjudicated 33% of deaths classified as ‘sudden cardiac death’ attributable to acute coronary findings (coronary thrombosis) whereas 37% of deaths originally classified as ‘progressive HF’ were reclassified as attributable to coronary thrombosis, although few were classed as ‘definite acute MI’.5

The risk of thrombo-embolic events (stroke, pulmonary, and peripheral thrombo-embolism) in patients with HF demonstrates little consistency in post-hoc analyses of large HF treatment trials of patients with HFrEF39⁻47(Table 2).

In general, these post-hoc analyses from HFrEF trials have reported a somewhat lower stroke incidence when compared with relatively smaller prospective observational studies (see Table 1) (Table 2). In published HF trials, annual stroke rates between 1.1% and 4.6% have been reported, but almost all of these analyses included some patients with AF.

In the Cooperative North Scandinavian Enalapril Survival Study (CONSENSUS), the highest absolute incidence of cerebrovascular events was observed, at 4.6% per year; however, this trial included patients with severe HF with high prevalence of concomitant AF.40 In an analysis of Vasodilator-Heart Failure Trials (V-HeFT I and II),41 the incidence of thrombo-embolic events in patients who were not receiving oral anticoagulation was 2.7% and 2.1% per year, respectively, in the V-HeFT I and II trials. However, ~15% of the patients were in AF, although surprisingly this condition was not found to be independently associated with an increase rate of thrombo-embolism.

In a recent analysis of the SCD-HeFT trial42 of patients with NYHA class II and III HF without AF, the authors reported that the incidence of thrombo-embolism (most were strokes) by 4 years was 4.0%, with 2.6% in patients randomized to amiodarone, 3.2% in patients randomized to an implantable cadioverter defibrillator (ICD), and 6.0% in patients randomized to placebo (approximate rates of 0.7, 0.8, and 1.5% per year, respectively). On multivariable analysis, hypertension (P = 0.021) and decreasing LVEF (HR 0.82, 95% CI 0.69–0.97 per 5% increase in EF) were significant predictors of thrombo-embolism. Interestingly, treatment with amiodarone or an ICD was a significant predictor of thrombo-embolism-free survival (HR vs. placebo 0.57, 95% CI 0.33–0.99 for ICD; 0.44, 95% CI 0.24–0.80 for amiodarone).42

In an analysis of the Survival and Ventricular Enlargement (SAVE) Trial,43 the overall risk of stroke was 8.1% at 5 years. The risk of stroke was found to be twice as high in patients with EF <28%. Every 5% decrease in EF was associated with an 18% increase in stroke risk. Unfortunately, in this study, the authors did not exclude patients with AF and only reported stroke events.

Al-Kadra et al.46 reviewed data on warfarin use in 6797 patients enrolled in the Studies of Left Ventricular Dysfunction (SOLVD) trial and found that the use of warfarin was independently associated with a significant reduction in all-cause mortality (adjusted HR 0.76, 95% CI 0.65–0.89, P = 0.0006) and in the risk of death or hospital admission for HF (HR 0.82, 95% CI 0.72–0.93, P = 0.0002). The risk reduction was not significantly influenced by the presence of AF, age, EF, NYHA functional class, or aetiology.

In a separate analysis from SOLVD,47 antiplatelet therapy use also significantly reduced mortality from all causes (adjusted HR 0.82, 95% CI 0.73–0.92, P = 0.0005) and reduced risk of death or hospital admission for HF (adjusted HR 0.81, 95% CI 0.74–0.89, P < 0.0001), but the association between antiplatelet therapy use and survival was not observed in the enalapril group. In another retrospective analysis of the SOLVD trials,48 the annual rate of thrombo-embolic events was 2.4% in women and 1.8% in men (after excluding patients with AF). Lower EF was associated with higher thrombo-embolic event rates in women but not in men, and this difference was mainly related to an increment of pulmonary embolism events. Interestingly, women showed a 53% increased risk of thrombo-embolic events for every 10% impairment in EF.

Amongst the more recent trials in the post-MI period that included patients with LVSD, the VALIANT trial reported a post-hoc analysis showing that the rate of stroke in the first 45 days was 0.94% (95% CI 0.78–1.09), whilst the cumulative rates were 2.33% (95% CI 2.08–2.58), 3.41% (95% CI 3.09–3.73), and 4.21% (95% CI 3.73–4.68) for the first, second, and third years, respectively.48 In the recent EMPHASIS-HF, the stroke rate was 1.5–1.9% and, taking into consideration 21-month follow-up, this gives a rate of stroke of ~1%/year.49 In the CORONA trial, where 92% were taking an angiotensin-converting enzyme (ACE) inhibitor/angiotensin receptor blocker (ARB) and 75% were on beta-blockers, non-fatal stroke incidence was 1.4–1.5%, although 35% of these patients had AF and 23% were on anticoagulation therapy.50

Limited data on stroke rates in patients with HFpEF are available from post-hoc analyses from large clinical trials focused on HFpEF. These data suggest that risk of a stroke event (and also mortality due to stroke) may be similar. For example, the I-PRESERVE study51 reported 147 stroke events in total, which gives an event rate of 0.8–0.9%/year; additionally, 9% of all deaths were due to stroke. Another example is the CHARM-Preserved study, where the rate of fatal and non-fatal stroke was entirely unrelated to EF (EF <22%, 1.2%; 23–32%, 1.4%; 33–42%, 1.4%; 43–52%, 1.3%; and >52%, 1.5%).52

Ongoing myocardial ischaemia may be a trigger in ~20% of cardiac arrest cases in hospitalized but stable patients with HF.53 This is reaffirmed by the ATLAS data discussed above, with acute (thrombotic) coronary findings at autopsy in many HF patients with sudden death.5Recognized MI per se is rather uncommon in HF trials.

Randomized controlled trials of antithrombotic therapy

There are a limited number of randomized clinical trials that have tested the efficacy of antithrombotic therapy in patients with HF in sinus rhythm. Table 3A shows those conducted before 1960 testing anticoagulant agents against no medication.54–56 These studies were performed in hospitalized patients with a high prevalence (up to 30%) of rheumatic valvular disease and AF. Randomization methods were probably biased and methods used to monitor the patients, and to determine patient inclusion and exclusion, were probably not as robust as in modern trials. Despite these limitations, all three studies favoured anticoagulant therapy in terms of a reduction in all-cause mortality and thrombo-embolic events (Table 3A). In the absence of accurate diagnostic imaging capabilities, most of the events were judged clinically or at autopsy, and, as such, strokes were not always diagnosed. In terms of bleeding, although minor bleeding episodes were reported in the active treatment arm, no significant increase in major bleeding was reported for the anticoagulated patients.

Table 3B reports randomized controlled trials performed well after 1961. There are four studies, three of which have been formally published.57–59 However, of these three trials, Warfarin/Aspirin Study in Heart Failure (WASH) was a pilot for the Warfarin and Antiplatelet Therapy in Chronic Heart failure (WATCH trial, and the subsequent WATCH and HEart failure Long-term Antithrombotic Study (HELAS) trials were stopped early due to slow recruitment rates. In the WASH study,57 279 patients were randomized to receive warfarin, aspirin, or no antithrombotic therapy and were followed-up for a mean (SD) of 27 (1) months to compare the rate of all-cause death, non-fatal MI, and cerebrovascular accident (CVA). There was no difference in the primary endpoint amongst the three treatment groups. However, those receiving aspirin were more likely to be hospitalized. Minor bleeding was higher among those receiving warfarin and aspirin, although there were few major haemorrhagic events overall.

The HELAS trial58 separated 197 patients according to the aetiology of their HF. The ischaemic cardiomyopathy group (n = 115) was randomized to receive aspirin or warfarin, while the non-ischaemic cardiomyopathy group (n = 82) was randomized to receive placebo or warfarin. Patients were followed up for a mean period of 18.5–21.9 months, depending on the group. No difference in the composite primary endpoint was observed. Major haemorrhage was seen among warfarin groups but not in other groups.

The WATCH trial59 is the largest published trial of antithrombotic therapy in HF patients to date; it assessed the rate of death, non-fatal stroke, and non-fatal MI in 1587 patients receiving aspirin, clopidogrel, or warfarin, who were followed up for a mean duration of 1.9 years. Of note, there was no placebo arm to truly assess ‘efficacy’ per se. There was no difference in the rate of the composite primary endpoint between treatment groups. However, the use of warfarin appeared to reduce the incidence of stroke compared with aspirin or clopidogrel. Major bleeding was higher with warfarin than with clopidogrel or aspirin. It was also noted that HF hospitalizations were significantly increased in subjects receiving aspirin.

The Warfarin versus Aspirin in patients with Reduced Cardiac Ejection Fraction (WARCEF) study is the only completed modern trial to date, and compares the efficacy of aspirin or warfarin on the primary endpoint of the first to occur of death, ischaemic stroke, or intracerebral haemorrhage in HF patients in sinus rhythm.60 This study is unique among trials of antithrombotic therapy in HF in sinus rhythm in that it employed a double-blind, double-dummy study design. All patients took real warfarin or aspirin, and dummy warfarin or aspirin; however, there was no control arm receiving no antithrombotic therapy, which precludes quantification of the effect of warfarin (or aspirin) vs. no treatment. All patients underwent international normalized ratio (INR) draw regularly, and sham INRs were generated for those on dummy warfarin (target INR 2.5–3.0). In WARCEF, 2305 subjects (mean age 62 years, 80% male, mean EF 25%) were enrolled in 11 countries, with final status known in 2245 (97.4%) after a mean of 3.5 years follow-up.

The primary results of WARCEF were first presented at the 2012 International Stroke Conference.61 The total follow-up time was 4045 patient-years in the warfarin group and 4033 patient-years in the aspirin group (range of 1–6 years per patient). The primary outcome was the first to occur of death, ischaemic stroke, or intracerebral haemorrhage. The main secondary endpoint adds intracranial haemorrhage, HF hospitalization, and MI to the primary endpoint.

Overall, there was no significant difference seen in the primary endpoint between the warfarin and aspirin groups (7.47% vs. 7.93%/year, respectively, HR 0.93, 95% CI 0.79–1.10, P = 0.40). However, there was a significant reduction in ischaemic stroke among those on warfarin vs. aspirin (0.72% vs. 1.36%/year, HR 0.52, 95% CI 0.33–0.82, P = 0.005). Additionally, there was some suggestion of a warfarin benefit for primary outcome for those followed for ≥4 years. For the main secondary outcome, which included primary endpoint outcomes plus MI and HF hospitalizations, there was no difference between the groups. Major haemorrhage was significantly higher with warfarin, although there were no significant differences in intracerebral or intracranial haemorrhage. In contrast to previous trials (i.e. WASH and WATCH), there was no evidence of increased HF hospitalizations in aspirin-treated patients.61

Clinical assessment/evaluation

Some HF groups at high thrombo-embolic risk merit specific clinical assessment and consideration for anticoagulation, although recommendations in previous clinical practice guidelines have varied in the level of detail62–64 (Table 5). In addition, anticoagulation may be considered for patients with right heart failure and pulmonary hypertension.

A high index of suspicion is prudent. Also, assessment for underlying paroxysmal AF may be needed, given the close relationship between AF burden and thrombo-embolism. While direct studies from HF populations are lacking, one systematic review concluded that AF may be detected in 1 in 20 cases presenting with acute ischaemic stroke, especially with prolonged electrocardiogram (ECG) monitoring (e.g. 7 day Holter monitoring, etc.).65 Thus, the harder one looks, the more likely one is to find AF.

Also, subclinical atrial tachyarrhythmias, without clinical AF, occur frequently in patients with pacemakers and can be associated with a significantly increased risk of ischaemic stroke or systemic thrombo-embolism.6 Detection of such arrhythmic events may be possible in HF patients with implantable electrical devices.

One ongoing multicentre, randomized trial of remote surveillance technology in patients (usually with reduced EF) with implanted dual-chamber cardiac resynchronization therapy defibrillator (CRT-D) devices is testing the hypothesis that initiation and withdrawal of oral anticoagulant therapy guided by continuous ambulatory monitoring of the atrial electrogram improves clinical outcomes by reducing the combined rate of stroke, systemic embolism, and major bleeding compared with conventional clinical management.66

Evaluation of bleeding risk is also mandatory as HF patients have a significant risk of bleeding on warfarin, especially when INR control is erratic due to liver congestion. Although bleeding risk scores have been developed and validated mainly in AF populations, they may also be applicable to HF patients in sinus rhythm to evaluate the potential risks of anticoagulation.67 It seem that using the well-validated HAS-BLED score68 may be recommended in these patients with careful follow-up if anticoagulant is prescribed.

Additionally it should be remembered that there is often a ‘high risk’ period for bleeding, following initiation of oral anticoagulation.69In elderly patients, the risk of major bleeding (and intracranial haemorrhage) with warfarin and aspirin may be similar.70 A recent systematic review and meta-analysis of randomized trials comparing warfarin and aspirin found a non-significant trend towards an increase in major bleeding risk in those randomized to warfarin [odds ratio (OR) 1.27, 95% CI 0.83–1.94].71 The pooled ORs for intracranial haemorrhage in patients treated with warfarin vs. aspirin was 1.64 (95% CI 0.71–3.78) and for extracranial major bleeding it was 1.03 (95% CI 0.61–1.75).71

Detection/imaging of intracardiac thrombus

On transthoracic echocardiography (TTE), intracardiac thrombi appear as dense, echogenic, heterogeneous, convex masses, with distinct margins, sessile or pedunculated, near thin dyskinetic ventricular segments or in the atrial appendages. Differential diagnoses include vegetations, tumours, devices, and artefacts, and intravenous agitated saline or transpulmonary contrast can contribute to the imaging diagnosis.72

Detection of cardiac thrombi is essential as anticoagulation may potentially reduce the cardioembolic risk, although the age of the thrombus and whether it is endothelized would influence its embolic potential.

About a third of all strokes may be cardioembolic. In one series of ischaemic stroke patients, transoesophageal echo (TOE) performed within 1 week of ischaemic stroke detected thrombus in approximately a quarter of cases.73 The yield of TTE is considerably lower than that of TOE, as TTE may be limited by suboptimal images (especially in obese patients), restricted field (e.g. in visualizing the apex or atria), and poor tissue characterization; of note, TTE is inferior to TOE for exploring posterior structures such as the left atrial appendage.74 Ischaemic stroke patients with a negative TTE result may still show potential cardiac sources of emboli in 50% of cases by contrast TOE.75

Cardioembolic sources include, in addition to visible thrombi, valve prostheses, catheter leads, central lines, mitral stenosis, patent foramen ovale, spontaneous echocontrast (swirling), reduced LV systolic or diastolic function, thin mobile mitral strands, and, perhaps, pulmonary vein ablation–isolation.75 In one study of 151 ischaemic stroke patients, those with either coronary artery disease, large strokes, AF, ECG evidence of ischaemia, or LVSD were more likely to have intracardiac thrombus on TOE. These factors may thus help direct stroke patients towards TOE.73 On the other hand, in stroke patients with no clinical cardiac disease and negative TTE, the yield of TOE for cardiac masses is considered to be very low, if not virtually nil.74

TOE is semi-invasive and more costly than TTE. Compared with echocardiography, magnetic resonance (MR) imaging or ultrafast computed tomography (CT) offer wide fields, are less operator dependent, can be performed in morbidly obese patients, have high spatial and temporal resolution, and allow tissue characterization.76,77 In one series, gadolinium-enhanced MR detected ventricular thrombi in 12/57 (21%) patients with ischaemic cardiomyopathy or prior MI—less than half of which were seen by TTE; moreover, TTE gave false-positive images in 3/57 (5%) patients.76 Thrombus on contrast-enhanced MR was related to larger end-diastolic volumes, lower ejection fractions, and LV aneurysm.76 CT is limited by potential nephrotoxicity of the intravenous contrast and by radiation exposure; MR is often not immediately available.77

Special situations

Role of aspirin in heart failure, bridging therapy, coronary interventions, presentation with acute coronary syndrome, devices (e.g. pacemakers, implantable cardioverter defibrillators), cardiac transplant.

Antiplatelet therapy with aspirin is common in patients with HF who are in sinus rhythm. Its use is based upon its general effect in the secondary prevention of atherothrombotic events. Prospective studies of antiplatelet agents in patients with HF in sinus rhythm are limited, although information is available in certain subgroups of patients, including those with AF and post-MI.92

Heart failure or moderate to severe LV systolic dysfunction is recognized as a risk factor for thrombo-embolism in patients with AF and, accordingly, most of the patients with this arrhythmia will require oral anticoagulation.93,94 Aspirin reduces the risk of death in the acute and early post-MI phase, although the mortality benefits are less significant during long-term treatment.95

As reviewed above, there is no available prospective evidence from long-term studies to recommend routine aspirin use as thromboprophylaxis in patients with HF who are in sinus rhythm. A possible interaction between aspirin and the ACE inhibitors may reduce the efficacy of the latter drugs and may account for more hospital admissions in those taking aspirin compared with warfarin.57,59 However, these findings have not been confirmed in the WARCEF study—the largest trial directly comparing warfarin with aspirin in optimally treated HFrEF population in sinus rhythm.61 Thus, HF per se is not a condition to recommend anticoagulation, but anticoagulants may potentially be considered for patients with presence of intracardiac thrombus, prolonged immobility/bed rest, AF, and previous systemic embolism or ischaemic stroke, as well as those with right heart failure and pulmonary hypertension.96

On the other hand, patients treated with drug-eluting coronary stents or presenting with an acute coronary syndrome will receive long-term dual antiplatelet therapy with aspirin plus a thienopyridine, usually clopidogrel, to reduce the risk of early and late stent thrombosis. In patients who require coronary angiography and are taking warfarin, common practice is to discontinue warfarin a few days prior to percutaneous coronary intervention to allow periprocedural INR levels to fall below therapeutic range (<2.0) and to start bridging therapy with unfractionated heparin or low molecular weight heparin, to cover the temporary discontinuation of oral anticoagulation, if the risk of thrombo-embolism is considered high.97 Recent recommendations suggest that uninterrupted anticoagulation with warfarin could replace heparin bridging in catheter interventions, with a favourable balance between bleeding and thrombotic complications.98–100

For HF patients requiring anticoagulation, options to reduce bleeding complications include shortening the duration of use of different antithrombotic drugs. Therefore, drug-eluting stents should be avoided or be strictly limited to specific clinical and/or anatomical situations, such as long lesions, small vessels, diabetes, etc.98–100 Patients who have chronic anticoagulation (usually with AF) presenting with an acute coronary syndrome and/or requiring percutaneous coronary intervention/stenting represent a complex management problem, and this has been reviewed in other consensus documents.98–100 Many anticoagulated patients with HF in sinus rhythm have stable coronary or carotid artery disease and/or peripheral arterial disease, and common practice is to treat such patients with warfarin plus one antiplatelet drug, usually aspirin. Adding aspirin to warfarin in such patients with stable vascular disease (whether coronary or peripheral artery disease) does not reduce the risk of stroke or vascular events (including MI), but substantially increases bleeding events.99–101

It may be necessary to interrupt oral anticoagulant therapy for elective implantation or replacement of a pacemaker or an ICD, although smaller procedures can often be performed without such interruption. In patients with HF in sinus rhythm, based on extrapolation from the annual rate of thrombo-embolism in patients with non-valvular AF, the consensus of the Task Force for the 2010 ESC guidelines suggests that anticoagulation may be interrupted temporarily for procedures that carry a risk of bleeding, such as ICD or pacemaker implantation, without substituting heparin.94 In patients at high thrombo-embolic risk (particularly those with prior stroke or transient ischaemic attack or systemic embolism), unfractionated heparin or low molecular weight heparin may be administered. It is possible that such operations can in part be performed without interruption of anticoagulation, as for vascular procedures.

In patients chronically anticoagulated with warfarin undergoing cardiac surgery, including heart transplantation, the risk of bleeding is likely to be increased when the INR is ≥1.5. Therefore, it is reasonable to reduce the INR to this level at the time of surgery. Several therapies are available for the reversal of oral anticoagulation, and these include oral or i.v. vitamin K, human fresh frozen plasma, prothrombin complex concentrates, and recombinant activated factor VII. Vitamin K alone is inappropriate if rapid normalization of the INR is required, because the onset of action is 4–6 h after i.v. administration and at least 24 h after oral administration. Therefore, when rapid reversal of warfarin is needed, vitamin K at doses of 2.5–5 mg should be administered i.v. in conjunction with other faster therapies.

Mechanical circulatory support (MCS) use is increasing during the last years, due to improvement of continuous-flow ventricular assist systems. After ventricular assist system implantation, a prothrombotic state has been observed, that is dependent on pump design and patient characteristics.102,103A ventricular assist system confers an increased risk for stroke, and the reported incidence varies from 3% to 47%.104 This hypercoagulable state requires commonly the use of individualized antithrombotic therapy, based on a combination of oral anticoagulation and antiplatelet therapy.105 Some authorities have recommended the use of thromboelastography to guide the monitoring of antithrombotic therapy in these patients.103

Role of oral anticoagulation (and antiplatelet therapy) for the reduction of stroke in heart failure with reduced ejection fraction and sinus rhythm

The risk reduction for stroke by warfarin vs. aspirin was very similar in the WATCH study59 (HR 0.58, 95% 0.25–1.3) and the WARCEF study (HR 0.52, 95% CI 0.33–0.82, P = 0.005).61 However, stroke was only a component of the primary endpoint in these studies, and neither of these studies showed a significant difference in the overall primary endpoint.

There appears to be a similar effect on stroke in these two studies which is of a similar magnitude to that seen for warfarin vs. aspirin in AF trials.106 However, this needs to be balanced against the risk of haemorrhage. A pre-specified secondary analysis for an endpoint of stroke alone is planned for WARCEF, and a combined analysis is planned pooling the WATCH and WARCEF data. These should provide more data on stroke risk reduction and take the risk of haemorrhage into account.

These two studies may still not offer a definitive answer to the question of whether antithrombotic therapy lowers the risk of thrombo-embolism in HF patients in sinus rhythm, as stroke alone was not a primary endpoint in either trial. It will also be important to determine the effect of warfarin in preventing stroke recurrence in HF.

Until more evidence becomes available, clinical decisions to treat patients with HF in sinus rhythm with anticoagulants must be made on a patient-by-patient basis, balancing the individual benefits against the risks of treatment, especially amongst high risk patients.

Alternative approaches that would probably be less useful would be to establish a prospective cohort study following patients with HF with and without warfarin for stroke events, as well as the establishment of a stroke registry in patients with HF and LVSD to compare stroke rates in larger numbers of HF patients with and without warfarin. One future large registry could prospectively evaluate the incidence of (silent or symptomatic) AF in HF patients in sinus rhythm. Such observational studies would have limitations, with the potential for major confounding factors.

There is also the perception that aspirin may be detrimental, and thus the CACHE (Clopidogrel versus Aspirin in Chronic HEart failure) trial (ISRCTN13415258)—which is a randomized, open-label, study that will compare the effects of aspirin and clopidogrel on outcomes in patients with chronic HF (http://controlled-trials.com/ISRCTN13415258/)—will test the hypothesis that when compared with clopidogrel, aspirin has an adverse effect on cardiovascular function in patients with HF including an increase in the symptoms of HF, reduced quality of life, an increase in hospitalizations, and a higher mortality.

Studies to identify a high stroke risk subgroup in patients with heart failure in sinus rhythm

To justify anticoagulation with warfarin, there is a view that the stroke rates in HF patients must be high enough (3–5% per year)107 since warfarin carries a risk of serious haemorrhage and necessitates frequent blood tests to monitor the anticoagulation effect. However, the use of newer safer oral anticoagulants108 that overcome the limitations of warfarin may shorten the ‘tipping point’ towards anticoagulation at stroke rates of ≥0.9%/year, at least in a Markov decision analysis model of patients with AF.109

Heart failure community studies3,107 have found low stroke rates, between 0.8% and 3.2% per year, and clinical studies also demonstrate similarly low annual stroke rates1 that may not justify routine anticoagulation in HF patients, even if warfarin has a stroke risk reduction effect similar to that in AF. A recent analysis of the REasons for Geographic and Racial Differences in Stroke (REGARDS) study was not able to define a subgroup of HF individuals with a rate of stroke high enough to warrant anticoagulation, but the stroke rates in HF were low.110

Further studies are needed to identify high risk subgroups in patients with HF and to establish a risk stratification scheme in HF similar to the CHADS₂ or CHA₂DS₂-VASc scores in AF.94 Similarly, there are established bleeding risk assessment scores for AF receiving anticoagulation, and we would need data to assess if similar scores, such as the HAS-BLED score recommended for use in AF patients, would be of similar value in predicting HF patients at risk of bleeding in whom antithrombotic therapy (whether aspirin or anticoagulants) is being considered.98

Confirmation of a high stroke rate within 30 days of onset of heart failure or of acute stroke in patients with heart failure

There is increasing evidence that stroke risk during the first 30 days after HF onset may be almost double the 5-year HF stroke risk, with a persisting smaller effect up to 6 months.2 This may also be true for the acute period after a stroke.24,25 These critical periods may be potential relative indications for anticoagulation in HF because of the higher stroke rates, and further data are needed on stroke incidence in these situations to determine if anticoagulation would be justified.

Assessment of current warfarin use in heart failure patients

Baseline data on use of warfarin from clinical HF studies suggest that up to 28% of HF patients in sinus rhythm are currently treated with warfarin (see above),46 but limited robust data exist. Since the indications for warfarin anticoagulation in HF in sinus rhythm are unclear, it is important to establish which patients are currently being treated.

Confirmation of low or normal ejection fraction as a risk factor for stroke in heart failure

Many clinicians use low ejection fraction as an indication to anticoagulate HF or cardiomyopathy patients, but there are inconsistencies in the data supporting low EF as a risk factor for stroke in HF; such data come largely from case–control studies or secondary analyses of clinical trials.111,112 Further data to confirm the risk of stroke, and whether the risk increases with decreasing EF, are needed, as well as to assess any interaction between low EF stroke rate and aetiology of LV dysfunction (ischaemic vs. non-ischaemic).

The majority of recent studies have focused on HF related to reduced EF. The impact of HF with normal or preserved EF on stroke and thrombo-embolism requires further attention, as do patients with preserved LV function but who have right heart failure with pulmonary hypertension. One recent analysis in hospitalized patients with AF did not find an independent contribution of EF to stroke and thrombo-embolism risk,113 but additional data in patients with HF in sinus rhythm are required.

The role of new oral anticoagulants in heart failure patients

Currently, new oral anticoagulants are available as alternatives to warfarin: these are in two broad classes, the oral direct thrombin inhibitors (e.g. dabigatran) and the oral Factor Xa inhibitors (e.g. rivaroxaban, apixaban).114

For example, apixaban has been compared with aspirin in patients with non-valvular AF who have refused warfarin or been deemed unsuitable for warfarin in the AVERROES trial.115 The latter was stopped early due to a clear superiority of apixaban over aspirin for the prevention of stroke and thrombo-embolism, with no significant difference between apixaban and aspirin for major bleeding or intracranial haemorrhage. Apixaban was also better tolerated than aspirin in AVERROES. The experience with apixaban in HF seems to be limited. Although in this trial symptoms of HF were present in 40% of patients, only 5% had evidence of reduced LVEF.115

Other agents (dabigatran and rivaroxaban) have only published phase III data in comparison with warfarin, but not aspirin; >60% of patients in ROCKET116 and one-third in RE-LY117 had symptoms of HF. Although both studies recruited patients in non-valvular AF, subgroup analyses revealed that there were no significant interactions between treatment with rivaroxaban or dabigatran and the presence of symptomatic HF.116,117 One network meta-analysis and indirect comparison study for dabigatran found that when compared with placebo, dabigatran etexilate 150 mg b.i.d. reduced the risk of any stroke (ischaemic and haemorrhagic) by 75%, ischaemic stroke by 77%, systemic embolism by 83%, and mortality by 36%. Dabigatran etexilate 150 mg b.i.d. also significantly reduced the risk of any stroke compared with aspirin monotherapy by 63% and aspirin plus clopidogrel by 61%.118 It is uncertain whether any of these new oral anticoagulants would be superior to aspirin (or warfarin, or no antithrombotic agent) in patients with HF in sinus rhythm. Further clinical trials are indicated.

Nonetheless, these new anticoagulant drugs are contraindicated in severe renal impairment (creatinine clearance <30 mL/min), and this is a concern in many patients with HF. There is no antidote or established reversal method for the anticoagulant action of these new drugs.119