Aspirin in Heart Failure: Don't Throw the Baby (Aspirin) Out with the Bathwater

Only a few years ago there was clearly a hype for giving a baby aspirin almost to any American male >50 years of age,1 and aspirin was predicated to be a fundamental component of the polypill.2 This was clearly an overstretching of indications, largely derived from the perceived ‘solid evidence’ for the use of aspirin in secondary cardiovascular prevention transferred to the much more controversial area of primary prevention. Because of its ease of use, and a perceived net clinical benefit—limited efficacy but better safety—compared with vitamin K antagonist oral anticoagulants (VKAs) such as warfarin, aspirin was also largely prescribed for stroke prevention in atrial fibrillation, essentially based on the result of one single trial,3 then impacting on a meta-analysis of eight trials.4 A number of facts and of opinions have more recently limited the enthusiasm for the use of aspirin. In primary cardiovascular prevention, several meta-analyses5-9 have led to inconclusive statements: in this setting, aspirin reduces the risk of myocardial infarction, does not apparently impact on death, and increases the risk of bleeding, including—possibly—a small increase in the risk of intracranial haemorrhage. In atrial fibrillation, where already in 2007 an important meta-analysis had concluded there was a clear inferiority of aspirin compared with VKAs,4 and where a trial even in the elderly had concluded there was an unfavourable net clinical benefit,10 clear recommendations against aspirin and in favour of oral anticoagulants were issued.11, 12 Now, in his editorial referring to the original paper by Chin et al.,13 Dr Cleland advocates not to use aspirin in heart failure—in any form of heart failure—based on the lack of evidence in this specific patient category.14 Such a recommendation needs to be carefully pondered, and this is the raison d'être for the present review, attempting to balance such an extreme view. Heart failure is a condition with diverse aetiologies. There is actually no agreed single classification system for the causes of heart failure, with much overlap between potential categories, and many patients having several different underlying pathologies—some of which non-cardiovascular—that conspire to cause the syndrome.15 It has been stated that identification of these diverse pathologies should be part of the diagnostic workup, as they may offer specific therapeutic opportunities.15 It is acknowledged that many patients with heart failure also have underlying coronary artery disease (CAD), often with a history of myocardial infarction or revascularization.15 Such conditions are allegedly atherothrombotic in their pathogenesis, and therefore at least partially preventable in relapses with the use of antithrombotic drugs, including aspirin. Indeed, it has been argued that both sudden death and death resulting from progressive heart failure may be caused by unrecognized atherothrombotic events.16 Heart failure is also associated with a hypercoagulable state: formation of left ventricular thrombus and cerebral embolism17, 18 have now, however, become relatively rare occurrences.19 As a result, there is a rationale for using antithrombotic agents to treat patients with ‘ischaemic’ heart failure who are in sinus rhythm. While the role of VKAs is now truly minor,20, 21 the possible role of one of the non-vitamin K antagonist oral anticoagulants is the subject of a current trial (COMMANDER).22 The use of aspirin is here much more widespread, and yet is the subject of controversy. The heterogeneity of aetiologies inevitably dilutes the potential benefit of a drug such as aspirin, only tackling one pathway—platelet cyclooxygenase (COX)-1—leading to one of the causes (atherothrombosis on the basis of CAD)23 of heart failure. In addition, heart failure has a morbidity and a mortality burden due to the disease itself, independent of its aetiologies.15 This inevitably limits the ability of demonstrating the prognostic impact of a treatment acting on one single pathogenetic event in the complex cascade leading to complications and outcomes. It therefore should come as no surprise that demonstration of benefits in such heterogeneous conditions is difficult, actually daunting, and never—as a matter of fact—indeed provided.24 In the secondary prevention of CAD, which also includes heart failure secondary to CAD, e.g. post-myocardial infarction, the latest Antithrombotic Trialists' (ATT) Collaboration meta-analysis estimated an effect of aspirin as an absolute reduction in serious vascular events (6.7% vs. 8.2% per year, P < 0.0001). This comes to the price of a non-significant increase in haemorrhagic stroke, but of reductions of about one-fifth in total stroke (2.08% vs. 2.54% per year, P = 0.002) and in coronary events (4.3% vs. 5.3% per year, P < 0.0001).5 In this secondary prevention setting, aspirin appears to be of substantial net benefit irrespective of age or sex, i.e. it would reduce non-fatal vascular events much more than it would increase major extracranial bleeds, and despite any adverse effects on cerebral haemorrhage.5 Overall, aspirin is also expected to reduce total vascular mortality.5, 25, 26 In 18 788 patients with a history of myocardial infarction in 12 trials, allocation to a mean duration of 27 months of antiplatelet therapy resulted in a highly significant reduction (2P < 0.00001) of 36 [standard error (SE) 5] fewer serious vascular events (death, myocardial infarction and stroke) per 1000 patients.26 Fewer vascular deaths (13 [standard deviation (SD) 5] per 1000; 2P < 0.005), as well as total deaths also occurred, together with a small, but still highly significant reduction in non-fatal stroke [6 (SD 2) per 1000; 2P = 0.0005].26 Thus, the evidence for aspirin benefit in patients with prior myocardial infarction, at time distance from the acute event, is solid, although derived—essentially because of study sizes—from a composite analysis of several trials. In the more acute setting of ‘unstable angina’, what we would now likely call ‘non-ST elevation myocardial infarction’, four (four!) placebo-controlled trials,27-30 with follow-up duration variable between 6 ± 3 days,28 12 weeks,27 3 months,30 and 2 years,29 demonstrated a reduction of major adverse cardiovascular events in the order of 40–60%, with statistical significance achieved in each of these single trials. The Second International Study of Infarct Survival (ISIS-2)31, 32 then demonstrated a significant reduction by aspirin of major cardiovascular events and of deaths. Actually, ISIS-2 had vascular mortality as the primary endpoint, and aspirin here achieved a highly significant 23% reduction in vascular mortality and a 50% reduction in re-infarction. One other trial, the Swedish Angina Pectoris Aspirin Trial (SAPAT), in patients with chronic stable angina,33 demonstrated a statistically significant 34% reduction in primary outcome events (myocardial infarction and sudden death; 95% confidence interval 24–49%; P = 0.003). The success of these demonstrations also considerably limit the relevance of the argument that plaque haemorrhage, a reported frequent precipitating cause of plaque rupture,34 is a prevailing mechanism for acute coronary syndromes, and therefore a counter-argument for the use of aspirin. If plaque haemorrhage were of such relevance, one could not explain the demonstration of efficacy of disparate antithrombotic therapies, both single (mostly aspirin—see above) and dual antiplatelet therapy,35 as well as anticoagulants,36 in preventing ischaemic events in CAD patients. The fact that no single early (before 1980) post-myocardial infarction trial in isolation in such a setting demonstrated a significant benefit on major cardiovascular events, and that such evidence was at that time derived from a meta-analysis of trials,37 later widely confirmed,5 should not be surprising. This is at variance from the setting of acute myocardial infarction and ‘unstable angina’, where multiple trials, as discussed above, demonstrated a significant reduction by aspirin of major cardiovascular events and also of deaths. With a similar proportional reduction of risk, the magnitude of the baseline risk is indeed the major determinant of the absolute benefit.23 Related to this is the sample size needed to demonstrate such an effect with statistical confidence. The accurate reader is referred to the first meta-analysis of aspirin trials by Peto, back in 1980.37 The landmark introduction of meta-analyses in the cardiovascular area overcame the sample size limitations of individual trials, and here specifically yielded the first unequivocal demonstration of the efficacy of aspirin after a myocardial infarction, indeed in patients all enrolled out of the acute phase.37 The significant reduction in later major cardiovascular events occurring also in stable angina patients, never having experienced an acute myocardial infarction,33 is also quite reassuring with regard to the possibility of confidently attributing to aspirin the ability to prevent major cardiovascular events in the long term, in the presence of CAD. These are the patients that eventually may develop post-infarction ‘ischaemic’ heart failure. Lifelong aspirin is therefore the current standard treatment for stable CAD,38 while waiting for the announced possibility that a small dose of one of the new anticoagulants, added on top of aspirin, or instead of aspirin,39 as currently tested in the COMPASS trial (NCT01776424),40 may change our practice. The words of Dr Cleland, ‘Adequate randomized controlled trials comparing aspirin to placebo in patients with heart failure are lacking but the few data that exist are not reassuring’, merit discussion. While a few trials investigated the relative benefits of aspirin and warfarin in heart failure,19, 20, 41, 42 the only real randomized controlled comparison of aspirin vs. placebo in heart failure independent of its aetiologies is the Warfarin/Aspirin Study in Heart failure (WASH),42 which was indeed inconclusive. The study randomized 279 patients with heterogeneous heart failure aetiologies, with a 627 patient-years exposure over a mean follow-up time of 27 months. The trial computed 49 major cardiovascular events, a composite of death, non-fatal myocardial infarction, or non-fatal stroke.42 By contrast, the previously cited first ‘historical’ meta-analysis leading to a consensus on the use of aspirin after myocardial infarction37 included a total of over 10 000 myocardial infarction patients randomized between aspirin and placebo control, with over 1000 deaths. Assuming that 50% of patients with heart failure have an ischaemic aetiology, and conservatively assuming that the clinical emergence of CAD (re-infarction) accounts for no more than one out of four deaths in patients with heart failure, then a trial in heart failure patients totalling at least 4000 deaths would be required to confirm or disprove the aspirin benefit in such patients. Is it therefore surprising that direct evidence for the use of aspirin in heart failure is inconclusive? And are there concerns ‘that aspirin might detract from the therapeutic benefits of agents that are known to improve outcome in heart failure, including [angiotensin converting enzyme] ACE inhibitors and beta-blockers, possibly by blocking prostaglandin production resulting in impaired vasodilatation, renal dysfunction, sodium and water retention, and hyponatraemia’?14 These possibilities have all been carefully scrutinized and overall excluded, at least as an argument not to use aspirin (see Patrono et al.43 and Eikelboom et al.44 for two comprehensive discussions on this specific topic). The Dagenais' meta-analysis of ACE inhibitors in stable vascular disease without left ventricular systolic dysfunction or heart failure45 apparently shows some heterogeneity in the risk reduction attributable to ACE inhibitors in the presence of antiplatelet agents, these latter including drugs different from aspirin, such as clopidogrel, and various doses of aspirin, not only the low doses here discussed. More important, however, that study still demonstrates benefits from ACE inhibitors in the presence of antiplatelet agents, with the 95% confidence intervals still not overlapping with the identity line. In agreement with this interpretation, the authors concluded that ‘The benefit [of ACE inhibitors] was shown in addition to other proven therapies such as antiplatelet agents, beta-blockers and lipid-lowering agents, indicating that combining these agents will provide the greatest clinical benefit’. Dagenais' meta-analysis45 included trials in which the use of aspirin and ACE inhibitors were not randomized. Such randomized data, conversely, exist for the Hypertension Optimal Treatment (HOT) trial for anti-hypertensive treatments, including ACE inhibitors.46 In the specific subanalysis in which 18 790 intensively treated hypertensive patients were randomized to either aspirin 75 mg daily or placebo for 3.8 years (with a 15% reduction in cardiovascular events and a 36% reduction in myocardial infarction in aspirin-treated patients), a reanalysis was conducted for the whole group of patients and for various subgroups, with particular attention to the possible effects of aspirin on blood pressure and renal function.47 In aspirin-treated and placebo-treated patients systolic blood pressure and diastolic blood pressure values achieved with antihypertensive treatment were superimposable, and the cardiovascular benefits of aspirin (found on ‘non-silent’—usually, however, clinically relevant—myocardial infarctions) were of the same magnitude in hypertensive patients receiving or not receiving ACE inhibitors.47 The authors concluded that no negative interaction occurs between ACE inhibition and the cardiovascular benefits of low-dose aspirin.47 Finally, in the 2012 authoritative review of antiplatelet agents by Eikelboom et al. on behalf of the American College of Chest Physicians Evidence-Based Clinical Practice Guidelines on Antithrombotic Therapy,44 the authors review the entire topic, and conclude that ‘Even though the results from these analyses cannot rule out the possibility of an interaction, they show unequivocally that even if aspirin is given, the addition of ACE inhibitor therapy produces substantial additional benefit in all major vascular outcomes. Therefore, in the absence of clear contraindications, concomitant use of aspirin and ACE inhibitors should be considered in all patients who are at high risk of major vascular events’. In conclusion, such interactions are either absent or of magnitude not annulling the effect of aspirin. Interactions have been conversely proved for non-steroidal anti-inflammatory drugs, and found on the one hand depending on the inhibition of renal COX-2, and on the other hand ruled out for low dose aspirin, which does not interfere with COX-2-dependent renal prostaglandin I2 (prostacyclin) synthesis.48 The recent paper by Chin is now reassuring as to the lack of an interaction with mineralocorticoid receptor antagonists,13 another mainstay in the therapy of heart failure. As in many cases in medicine when a new treatment comes to the horizon, the history of aspirin in heart disease has gone through phases of ‘skepticism, enthusiasm, disillusionment, and panic’.49 In the early '70s, six trials had been performed post myocardial infarction, with varying aspirin doses, leading to a chaotic picture of inconclusive trends. The only way to make sense of trials each inadequate in terms of sample size was to submit them to the newborn tool of a meta-analysis.37 This was subsequently confirmed by an increasing number of trials jointly analysed in the various publications of the Antiplatelet/Antithrombotic Trialists' Collaboration.5, 25, 26 The ISIS-2 landmark trial, in cases of ‘suspected’ acute myocardial infarction, definitely proved the use of aspirin in the acute phase of a myocardial infarction.31, 32 This was the phase of trumpeting enthusiasm for the use of aspirin. Evidence was, and overall still is, inconclusive in the area of primary prevention, which is, however, an extremely heterogeneous conundrum with a highly variable risk of major adverse cardiac events.50 We have argued that an assessment of the baseline cardiovascular risk is still the best tool to choose in whom to give aspirin,50 but others have argued against this.51 At the same time, skepticism had been cast on the use of aspirin as an antithrombotic strategy in atrial fibrillation, for the clear superiority of anticoagulants in general in stroke prevention,11, 52 and also because of the net clinical benefit associated with VKA use also in the elderly.10 This has led to a general disillusionment, and also perceived disfavour, for the use of a drug that is life-saving when given in the appropriate patient context, complementary to other risk reduction strategies such as statins, ACE inhibitors, and angiotensin receptor blockers,23 and now argued not to interfere with the action of mineralocorticosteroid receptor antagonists.13 In current days, new discoveries and evidence on the effect of aspirin on cancer,53 even at low doses specifically tackling platelet COX-1,54, 55 are now revamping again the interest for the use of the drug in primary cardiovascular prevention and beyond (Figure 1). We are not yet at the point of reaching unanimous consensus, but still international guidelines are aligned in recommending the use of aspirin lifelong in secondary CAD prevention,38 including when this complicates with heart failure. This is not based on direct evidence in this specific setting, but on logical considerations, the same that prompt the use of a parachute in jumping from an airplane even in the absence of a formal trial. The 2013 ACCF/AHA guideline for the management of heart failure states that the benefit afforded by low-dose aspirin in patients with systolic (reduced ejection fraction) heart failure but no previous myocardial infarction or known CAD (or specifically in patients proved to be free of CAD) remains unknown.56 This would imply a continuing recommendation for aspirin in patients with heart failure with proven CAD. Similarly, the 2016 ESC guidelines on heart failure state: ‘…there is no evidence on the benefits of antiplatelet drugs (including acetylsalicylic acid) in patients with heart failure without accompanying CAD, whereas there is a substantial risk of gastrointestinal bleeding, particularly in elderly subjects, related with this treatment’.15 This also implies that aspirin should be given in patients with heart failure with accompanying CAD, although its use should be compounded (as always) by an adequate evaluation of the expected benefits and risks. I continue to practice this way, not negating the potential benefits of aspirin in patients with heart failure with concomitant CAD. R.D.C. expresses gratitude to Professor Carlo Patrono for insightful discussions on this topic. Conflict of interest: R.D.C. reports honoraria and speaker fees from Bayer Healthcare in relation to aspirin use and misuse.