Stroke Prevention by Low-dose Anticoagulation in Patients With Heart Failure and Sinus Rhythm

Is It Worth the Effort?

Aldo P. Maggioni; Felicita Andreotti


Eur Heart J. 2019;40(44):3602-3604. 

Oral anticoagulants are recommended to prevent arterial thrombo-embolic events in patients with atrial fibrillation, left ventricular thrombus, or mechanical valves, as their use is associated with reduced rates of ischaemic stroke, peripheral embolism, and other atherothrombotic events such as myocardial infarction and acute limb ischaemia.[1] Patients in sinus rhythm with heart failure and reduced ejection fraction (HFrEF) have neuroendocrine and pro-coagulant alterations,[2] structural and functional cardiovascular abnormalities, and low systemic output, which that predispose to arterial thromboembolism;[1] haemostatic alterations may also contribute to the adverse spiral of pathophysiological events in HFrEF.

On the above premises, the COMMANDER trial[3] tested whether low-dose rivaroxaban [2.5 mg b.i.d. (twice daily)] vs. placebo could improve outcomes in 5022 patients with worsening chronic HFrEF of ischaemic aetiology and sinus rhythm. Most patients were on antiplatelet therapy. After 21 months, the primary outcome of death, myocardial infarction, and stroke occurred in 25.0 and 26.2% (P=0.27), respectively;[3] fatal or critical space bleeding did not differ, but International Society on Thrombosis and Haemostasis (ISTH) major bleeding (that additionally includes haemoglobin drops ≥2 g/dl or transfusions ≥2 units) increased from 1.21 to 2.01% with rivaroxaban (P=0.003).[3]

Although COMMANDER showed no significant differences between treatments in the primary efficacy outcome, a post hoc analysis did indicate significant reductions of thrombo-embolic events with rivaroxaban, from 15.5 to 13.1% when sudden/unwitnessed deaths were included (P=0.01) or from 7.6 to 6.1% when sudden/unwitnessed deaths were excluded (P=0.04).[4] Overall, the findings suggested that, in a worsening HFrEF population in sinus rhythm, anticoagulation works on what it is mainly meant to do, i.e. prevention of thrombo-embolic events, but with little effect on deaths unrelated to thromboembolism, such as those caused by pump failure or arrhythmia.

In this issue of the European Heart Journal, Mehra et al. conducted a further detailed post hoc analysis of the COMMANDER population, focusing on the incidence, timing, type, severity, and predictors of stroke and transient ischaemic attacks (TIAs); approximately one-half of these events were fatal or disabling.[5] With low-dose rivaroxaban, the occurrence of ischaemic stroke and TIAs dropped by 32%, from 3.5 to 2.4% (P=0.02). The paper by Mehra et al. also describes the rate of occurrence and the temporal trends of cerebrovascular events. In patients with a CHA2DS2-VASc score >4, the authors show that the number needed to treat (NTT) for rivaroxaban 2.5 mg b.i.d. is 96 per year. The report by Mehra et al. has several limitations, as it is a post hoc analysis of a trial with neutral results; TIA events were not centrally adjudicated by an independent committee, and patients with strokes <3 months were not included, possibly leading to an underestimation of the rate of occurrence of further cerebrovascular events. Nonetheless, the findings support the effectiveness of low-dose rivaroxaban in preventing the most feared thrombo-embolic event, i.e. cerebral infarction and its harbinger, cerebral ischaemia.

The results reported by Mehra et al. go along the lines of two previous trials, WARCEF[6] and WATCH,[7] that compared warfarin against antiplatelet therapy in patients with HFrEF and sinus rhythm. WARCEF tested full-intensity warfarin against aspirin 325 mg daily in 2305 patients over 3.5 years; it found that anticoagulation did not significantly influence the primary outcome of death and stroke, but reduced ischaemic stroke rates from 1.36 to 0.76% per year (P=0.005), while increasing ISTH major (but not intracranial) bleeds from 0.87 to 1.78% per year. WATCH tested full-intensity warfarin against aspirin 162 mg daily or clopidogrel 75 mg daily in 1587 patients over a median of 21 months; it also found that anticoagulation vs. single antiplatelet therapy did not significantly influence the primary outcome of death, myocardial infarction, and stroke, but reduced all stroke rates from 2.3% with aspirin or clopidogrel to 0.6% with warfarin (P=0.016), while increasing modified ISTH major (but not intracranial) bleeds from 2.1% with clopidogrel to 5.2% with warfarin (P=0.007).

Now, if, despite the above-mentioned limitations, the favourable message from the paper by Mehra et al. can be reasonably accepted, based on plausibility and on supportive data from previous studies, a crucial question is: to what extent do stroke and TIAs contribute to the overall burden of adverse events in a clinical condition such as HFrEF? Is the potential to prevent cerebrovascular events going to modify the natural and clinical history of patients who generally die, or are strongly limited in terms of quality of life, owing to their progression of left ventricular dysfunction and failure? Figure 1 (panel A) reports the incidence of strokes and all-cause mortality in the control group of trials of patients with heart failure.[3,6–10] All these studies show a consistent finding: fatal and non-fatal stroke events are numerically a small minority with respect to all-cause death. For this reason, an oral anticoagulant, even if effective on the events for which it is prescribed, cannot have a major impact on outcome measures that include mortality for the clinical condition of HFrEF. Specific trials testing anticoagulant drugs for HFrEF in sinus rhythm should define thrombo-embolic or stroke events as their primary endpoint, with cardiovascular or all-cause mortality as secondary endpoints. This is likely the only way to reliably demonstrate the efficacy of oral anticoagulants in this clinical setting.

Figure 1.

(A) Rates of ischaemic stroke and all-cause mortality in the control group of trials in patients with heart failure. (B) Oral anticoagulants vs. control in trials in patients with heart failure in sinus rhythm. CAD, coronary artery disease; EF, ejection fraction; FU, follow-up; HF, heart failure; HFrEF, heart failure with reduced ejection fraction; OAC, oral anticoagulant.

Another field, very scarcely investigated, is that of patients with heart failure and preserved ejection fraction (HFpEF). They account for nearly one-half of all patients with heart failure in real-world clinical practice, are generally older than those with HFrEF, and are more frequently hypertensive. Owing to these peculiar clinical characteristics, the stroke occurrence in patients with HFpEF could be more frequent than in those with HFrEF. In this population of patients, specific studies to assess the role of oral anticoagulation are also dearly needed.

Is there a role for other antithrombotic drugs, i.e. antiplatelet agents, in the clinical context of HFrEF with sinus rhythm? In addition to WARCEF and WATCH, two smaller trials, WASH and HELAS, compared warfarin vs. aspirin or placebo over 627 and 312 patient-years.[11,12] In these two trials combined, there were nine strokes, with little power to analyse any effect of treatment.[11,12] Overall, the available evidence suggests superiority of anticoagulation over antiplatelets or placebo for stroke prevention in patients with HFrEF and sinus rhythm (Figure 1, panel B), at the price of increased extracranial, but not intracranial, major bleeds.

Clinical decisions, even in the absence of evidence-based information, need to be taken daily when a patient with HFrEF is in front of us either in-hospital or as an outpatient. If the patient presents with atrial fibrillation or with images of left ventricular thrombus, there is a clear indication to treat him/her with full-dose oral anticoagulation. If the patient is in sinus rhythm, the CHA2DS2-VASc score might be considered, since it works in predicting the risk of stroke even in these patients, as suggested by Mehra et al.,[5] specifically if the score is >4. If the risk of bleeding is not very high, might rivaroxaban 2.5 mg b.i.d. become a possible future therapeutic option, considering the very high mortality and severe disability that a stroke event can determine in this population of patients?

In conclusion, despite relevant limitations, the paper of Mehra et al. explores a topic generally overlooked by trialists working in the field of heart failure, providing thought-provoking observations, interesting hypotheses for new studies, and potentially useful suggestions for the management of patients in daily practice.