Antithrombotic therapy in heart failure and sinus rhythm: the ongoing search for a better match of patients to therapy

This article refers to ‘Plasma D-dimer concentrations predicting stroke risk and rivaroxaban benefit in patients with heart failure and sinus rhythm: an analysis from the COMMANDER-HF trial’ by J.P. Ferreira et al., published in this issue on pages 648–656. Patients with ischaemic heart disease remain at risk of recurrent ischaemic events despite receiving antiplatelet therapy. Thrombocardiology is a fast moving field, and various combinations of antithrombotic drugs instead of the traditional ‘aspirin monotherapy approach’, have provided incremental ischaemic benefits but often also higher bleeding liability.1, 2 The advent of newer direct-acting oral anticoagulants, offering an improved risk–benefit profile compared with vitamin K antagonists, has renewed the interest for long-term anticoagulation and fostered the investigation of a dual-pathway inhibition strategy, consisting of the combination of a single antiplatelet and an anticoagulant agent.3 However, patients who suffered from ischaemic heart disease, and, yet at different risks of recurrent ischaemic events, such as those with recent acute coronary syndrome (ACS), chronic coronary artery disease (CAD), or with worsening heart failure (HF) and reduced ejection fraction (HFrEF) may have different responses to drug combinations. The association of low-dose rivaroxaban (2.5 mg bid and 5 mg bid) with mainly dual antiplatelet therapy was tested in 15 526 patients with ACS in the ATLAS ACS 2–TIMI-51 trial,4 and demonstrated significant reduction of the composite of cardiovascular death, myocardial infarction (MI), or stroke, in the rivaroxaban groups vs. placebo (8.9% vs. 10.7%, P = 0.008), but also greater major bleeding and intracranial bleeding risks.4 ACS patients with a history of HF derived greater benefit from rivaroxaban treatment.5 The COMPASS trial,6 enrolled patients with chronic CAD or peripheral artery disease, but without recently decompensated or severe HF (defined by baseline ejection fraction of ≤30% or New York Heart Association functional class III or IV HF), and showed that a low dose (2.5 mg bid) of rivaroxaban with aspirin was also associated with lower cardiovascular death, MI, or stroke rates than aspirin alone (4.1% vs. 5.4%, P < 0.001), and consistent higher major but not fatal or intracranial bleeding risks.6 In the COMMANDER-HF trial,7 5022 patients with HFrEF with a recent episode (within 21 days) of worsening HF, established CAD and without known atrial fibrillation, were randomized to low-dose rivaroxaban (2.5 mg bid) or placebo. Aspirin alone, or in combination with a thienopyridine, was taken by 93.1% of the patients, with 34.8% of the patients being on dual antiplatelet therapy at baseline. Over a median follow-up period of 21.1 months, rivaroxaban did not reduce the incidence of the primary outcome of death from any cause, MI, or stroke compared to placebo (25.0% vs. 26.2%, P = 0.27) but did result in higher risk of International Society on Thrombosis and Haemostasis (ISTH) defined major bleeding.7 These results highlight a gradient of ischaemic benefits for rivaroxaban-based anticoagulation depending on the clinical presentation, which seems clinically meaningful and negligible in patients with recent ACS and CAD-HFrEF, respectively. Although patients enrolled across these trials had to have established CAD or an atherosclerotic vascular disease and sinus rhythm, thrombin-mediated events were most likely not a major driver for death or rehospitalization among HFrEF patients with recent decompensation, potentially explaining the study findings. Conversely, pump failure or arrhythmic deaths (which were retrospectively not substantially impacted by treatment with rivaroxaban), contributed to a substantial proportion of all-cause mortality. These results are consistent with prior observations among older randomized trials in HF and sinus rhythm patients, in which warfarin compared with aspirin showed no mortality benefit and no significant difference in non-fatal MI, but significant major bleeding risks.8 The WARCEF trial,9 the previous largest trial enrolling patients (n = 2305) with HFrEF and sinus rhythm, showed no significant difference in first event of ischaemic stroke, intracerebral haemorrhage, or all-cause death with warfarin compared to aspirin. However, warfarin showed a significant reduction in the rate of ischaemic stroke throughout the follow-up period, a benefit suggested also in the smaller WATCH trial.10 In this perspective, it remains worth noting that rivaroxaban 2.5 mg bid reduced stroke risk in both COMPASS and COMMANDER-HF trials.6, 7 Therefore, the potential identification of patients who may derive the greatest benefit from a more aggressive antithrombotic regimen is relevant for a tailored and optimized therapy among HF patients. In this issue of the Journal, Ferreira et al.11 undertook a post-hoc analysis of the COMMANDER-HF trial assessing the association between baseline plasma D-dimer concentrations and the risks of stroke and/or other fatal and non-fatal athero-thrombotic events as well as bleeding. The authors included 4107 patients in whom D-dimer was assessed (i.e. 82% of the entire study population) and stratified the population into D-dimer tertiles: (i) ≤255 ng/mL; (ii) 256–515 ng/mL; (iii) >515 ng/mL. Elevated D-dimer concentrations were independently associated with higher rates of all-cause, cardiovascular and sudden death, HF rehospitalization, stroke of any cause, ischaemic stroke and deep vein thrombosis/pulmonary embolism but not MI. These results confirm the prognostic role of D-dimer concentrations for adverse outcomes.12 D-dimer, a product of fibrin degradation which results from ordered breakdown of thrombi by the fibrinolytic system, may serve as a marker of activation of coagulation and fibrinolysis. The clinical utility of D-dimer measurement has been established for the diagnosis of deep vein thrombosis and disseminated intravascular coagulation. However, D-dimer measurements have increasingly been studied and associated with multiple diseases and clinical scenarios (vascular events, cancer incidence events, and multiple cause-specific mortalities),12 making it a rather unspecific marker. The unresolved issue remains what drives D-dimer elevation in these patients. Moreover, in this analysis, D-dimer levels were measured at baseline after an episode of worsening HF and yet they were markers of events occurring at long-term follow-up. Most importantly, Ferreira et al.11 showed the effect of rivaroxaban was similar across D-dimer tertiles for all outcomes except for overall and ischaemic stroke whereas only patients with higher plasma D-dimer concentrations (above ≈500 ng/mL) appeared to benefit from rivaroxaban for stroke reduction (interaction P = 0.008). These results are of particular interest but should be interpreted cautiously because they were obtained from a non-pre-specified analysis in the context of a negative trial, and by exploring a multiplicity of outcomes. The favourable hazard ratio reported for rivaroxaban on stroke in the highest D-dimer tertile (and the significant interaction with treatment) is strong but derived from a relative small number of stroke events observed in this subgroup. Moreover, while patients within the third tertile derived the greatest absolute and relative stroke reduction from treatment, the absence of a dose–response association (hazard ratio first increases then decreases with increasing D-dimer tertiles: tertile 1 = 1.16, tertile 2 = 1.45, and tertile 3 = 0.36; interaction P = 0.008) suggests that further investigations are needed for a better understanding of the mechanism of stroke prevention. Finally, it should be noted that stroke events were investigator-reported and did not undergo central adjudication. Despite the above-mentioned limitations, this hypothesis-generating analysis sets the stage for a clinical trial exploring the safety and efficacy of rivaroxaban in HF patients with sinus rhythm and high D-dimer levels (Figure 1). Even more so, considering that rivaroxaban effect on bleeding was not influenced by plasma D-dimer concentrations; thus, potentially targeting an HF population with low risk to benefit ratio. Heart failure itself is a persistent risk factor for ischaemic stroke13 and emerging data suggest a high incidence of silent atrial fibrillation among HF and sinus rhythm patients with cardiac implantable electronic devices.14 On the other hand, stroke events are not a major cause of morbidity and mortality in patients with HFrHF with recent worsening of HF, as shown in COMMANDER-HF, where they contributed to less than 10% of the composite primary endpoint. Should all patients with HFrEF undergo routine screening for D-dimer levels for the identification of the ideal oral anticoagulation candidates? And if so, for how long should the treatment be considered? Of course, these critical questions can only be answered by dedicated randomized trials. Despite all progresses made over the last years in HF management,15 a dedicated anti-thrombotic therapy for HFrEF patients and sinus rhythm lacks any supporting evidence, which may be particularly critical for recently decompensated HF or advanced HFrEF. As of today, the liberal use of anticoagulation in this challenging population cannot be recommended, the optimal anti-thrombotic treatment remains an evidence-free zone and the search for a better match of patients to therapy continues. Conflict of interest: E.F. has nothing to disclose. G.S. reports personal fees from Biotronik, Boston Scientific, AstraZeneca and Novartis, outside the submitted work. M.G. reports personal fees from AstraZeneca, Alvimedica/CID, Abbott Vascular, Bayer, Biotronik, CoreFLOW, IDORSIA Pharmaceuticals Ltd, Universität Basel - Dept. Klinische Forschung, Bristol-Myers Squibb, Chiesi, Medscape, grants and personal fees from Terumo, Daiichi-Sankyo, outside the submitted work.