Antiplatelet agents in clinical practice and their haemorrhagic risk

Introduction

Platelets are the main effectors of primary haemostasis but also of arterial thrombosis. This dual role is at the origin of the inextricable Gordian knot of the last 20 years’ research on platelet pharmacology, i.e. the apparently unresolvable middle way between the Scylla’s cliffs of thrombosis and the Charybdis’ shallows of bleeding. Expanding knowledge on the mechanisms regulating the participation of platelets in haemostasis and thrombosis has led to great progress in antiplatelet therapy and may ultimately allow the bottleneck of current therapies to be overcome¹.

Platelets in haemostasis and thrombosis

Platelets circulate in a resting, inactive state in intact blood vessels, but they interact very rapidly and selectively with exposed subendothelial surfaces and functionally altered endothelial cells². The first interaction between platelets and the exposed subendothelial surface is mediated by von Willebrand factor (VWF), a high-molecular weight multimer protein synthesised and released by endothelial cells but also contained in and released by circulating blood platelets. This factor binds on one side to subendothelial collagen, via its A3 domain, and on the other side to platelet GPIb/IX/V, via its A1 domain, thus slowing-down the movement of platelets along the damaged vessel surface. Subsequently, platelet collagen receptors (GPIa/IIa and GPVI) come into contact with subendothelial collagen fibres leading to the complete arrest of platelets on the damaged vessel wall and to their activation. Upon activation platelets release arachidonic acid, contained in their membrane phospholipids, and produce, through a reaction catalysed by the enzyme cycloxygenase-1 (COX-1), thromboxane A₂ (TxA₂), a powerful platelet agonist that acts on specific receptors on the platelet surface (TP receptors). Small amounts of other products of COX-1, such as prostaglandin (PG)E₂ or PGG₂-H₂, may also contribute to platelet activation^2–4. Moreover, platelets release their granular contents, including ADP, a platelet agonist acting on specific receptors on the platelet surface (P2Y1 and P2Y12), and a number of other substances contributing to platelet activation (serotonin, matrix metalloproteinase [MMP]-2 and -1, etc.). The local formation of thrombin, triggered by the exposure of tissue factor and facilitated by the expression of a procoagulant surface on activated platelets, strongly contributes to the extension of the initial platelet activation process.

The pathological extension of platelet activation at a site of vascular damage may lead to thrombus formation, and occurs when either the mechanisms designed to prevent unwarranted extension of platelet activation are impaired^5–7 or when an accumulation of substances favouring platelet activation takes place².

The targets of currently used antiplatelet agents are platelet COX-1 (by aspirin), thus blocking the formation of TxA₂, and the platelet ADP receptor P2Y12 (by clopidogrel, prasugrel and ticagrelor), thus blocking the amplification loop that follows the initial platelet activation response².

Antiplatelet therapy: evidence of efficacy

Antiplatelet agents are the cornerstone of treatment of patients with a previous ischaemic cardiovascular event, i.e. in secondary prevention. In particular, treatment with aspirin prevents the occurrence of a new major cardiovascular event in approximately one fifth to one quarter of patients with a previous myocardial infarction, a previous stroke or transient ischaemic attack and in patients in other high-risk categories with a previous ischaemic event⁸.

In terms of absolute reduction of ischaemic events, antiplatelet therapy prevents 36 major cardiovascular events every 1,000 patients with a previous myocardial infarction treated for 27 months, 38 every 1,000 patients with an acute myocardial infarction treated for 1 month, 36 every 1,000 patients with a previous stroke or transient ischaemic attack treated for 29 months, and 22 every 1,000 patients with other high-risk conditions treated for 22 months⁸. The very clear advantage provided by antiplatelet therapy in terms of reduction of cardiovascular morbidity and mortality must be balanced against the weight of adverse events, and in particular of the risk of bleeding. In secondary prevention this balance is clearly favourable, with an evident predominance of the benefit (5 to 10% absolute risk reduction of an ischaemic event against a 0.1 to 0.5% absolute increase of major bleeding)⁹. In contrast, this balance is no longer advantageous in the setting of primary prevention, i.e. in patients with a relatively low cardiovascular risk, in whom the absolute reduction of ischaemic events would be about only twice the absolute increase in bleeding, irrespective of the age and sex of the population treated⁹.

By far the most frequent haemorrhagic complication associated with aspirin is gastrointestinal bleeding. In part this is due to the very mechanism of action of aspirin, i.e. blockade of the COX-1 enzyme which in the stomach plays a protective role against ulcers by producing gastroprotective PGE₂¹⁰. This detrimental effect of aspirin is dose-dependent, and in fact cumulative analysis of aspirin trials shows that the relative risk of an upper gastrointestinal bleed in people taking aspirin rises from 2.2- to almost 4-fold that of non-aspirin users in the range of doses between 75 and 300 mg/day¹¹ (Figure 1). This may have a large impact when projected to the huge population of patients currently taking aspirin. It is estimated that around 50 million patients take aspirin daily in the USA, and that 40% of these are prescribed a dose ≥300 mg/day. Extrapolating the above-mentioned gastrointestinal bleeding data with different aspirin doses to this large population implies that if all patients took a dose of aspirin ≤100 mg/day, excess gastrointestinal bleeding would be reduced by 900,000 cases/year¹².

Figure 1.

Increase of gastrointestinal (GI) bleeding in patients treated with different antithrombotic regimens. A) Dose-dependent increase in GI bleeding associated with aspirin use. B) Increase of GI bleeding associated with low-dose aspirin compared with other antiplatelet agents or with oral anticoagulants. RR = relative risk as compared with age- and sex-matched untreated controls. Other antiplatelet: ticlopidine, clopidogrel, trifunisal (modified from references 11,14). 95% CI =95% confidence interval.

The dose-dependency of major adverse bleeding events is also evident when aspirin is used in combination with clopidogrel, as observed in the CURE trial¹³. However part of the aspirin-related gastrointestinal bleeding is independent of its specific mechanism of action and is consequent to the mild impairment of haemostasis produced by antiplatelet treatment. Indeed, not only low-dose aspirin but also clopidogrel, which does not have any intrinsic gastrotoxicity, and even warfarin increase the relative risk of gastrointestinal bleeding to approximately the same extent (relative risk around 2–2.2 fold) (Figure 1), suggesting that in a consistent fraction of the population of patients undergoing antithrombotic therapy a pre-existing, clinically unnoticed, gastrointestinal ulcer starts to bleed when haemostasis is pharmacologically impaired^11,12,14.

As a general rule, antiplatelet therapy-related major bleeding becomes greater the stronger the platelet inhibition that is attained. This is evident when comparing the combination of aspirin plus clopidogrel with aspirin alone^15,16, aspirin plus prasugrel versus aspirin plus clopidogrel¹⁷ or aspirin plus ticagrelor versus aspirin plus clopidogrel¹⁸. Whether this is an unavoidable consequence of enhanced platelet inhibition or whether a magic bullet that can prevent thrombosis without enhancing bleeding exists is an important unsolved issue¹⁹.

Predictors of bleeding during antiplatelet therapy

One possible way out from the thrombosis/bleeding dualism is offered by the identification of those patients who are more likely to bleed when treated with antiplatelet therapy. Several studies have tried to define independent predictors of major bleeding in patients with acute coronary syndromes (ACS) treated with aggressive antithrombotic therapy and found that some clinical, personal case history, or laboratory parameters are consistently associated with an increased rate of bleeding, such as advanced age (>75 years), female gender, renal function impairment, anaemia, history of bleeding, low body weight. Low body weight (<60 kg), advanced age (>75 years) and a previous stroke or transient ischaemic attack were predictors of intracranial haemorrhage in the TRITON-TIMI 38 trial in prasugrel-treated patients¹⁷.

Negative prognostic value of bleeding during antiplatelet therapy

Major bleeding during antiplatelet therapy is not only dangerous for its immediate consequences but also has a strongly predictive value for long-term mortality. This was first reported after an analysis of data from three large clinical studies, the OASIS registry, the OASIS-2 and CURE trials, involving over 34,000 patients, which showed that patients undergoing major bleeding during their stay in hospital have a 5-fold higher incidence of death during the first 30 days and a 1.5-fold higher incidence of death between 30 days and 6 months²⁰. A stepwise increase in 30-day mortality depending on bleeding severity was also reported, with a strikingly higher mortality in patients with severe bleeding and a non-significant increase in patients with moderate or mild bleeding²¹. Moreover, in-hospital death rates in patients hospitalised for an ACS were consistently higher in bleeders than in non-bleeders across the whole spectrum of coronary events (patients with unstable angina, non-ST-elevated myocardial infarction, ST elevated myocardial infarction [STEMI]) in patients from the GRACE registry²². Major bleeding had a similar strength of association with mortality at 1 year as had myocardial infarction in patients with an ACS in the ACUITY trial²³. Further, strong proof of the negative prognostic value of bleeding in patients on antithrombotic therapy comes from the observation that therapeutic strategies associated with less bleeding generate a higher net therapeutic benefit. In the HORIZONS-AMI trial in 3,602 STEMI patients undergoing percutaneous coronary intervention (PCI), treatment with bivalirudin alone was associated with a strikingly lower major bleeding rate as compared with treatment with heparin plus a GPIIb/IIIa antagonist. Despite similar major ischaemic cardiovascular events rates in the two treatment groups, net clinical benefit and cardiac mortality were strikingly and significantly lower in the bivalirudin group²⁴ and the benefit persisted over a 3-year follow-up (HR in bivalirudin-treated patients at 3 years: 0.56, 95% CI, 0.4 to 0.8, P =0.001)²⁵.

Bleeding and enhanced ischaemic cardiovascular risk: mechanisms

Whether the association between major bleeding and increased mortality is merely an association based on shared risk factors or whether there is a cause-and-effect relationship remains to be fully established²⁶. Several possible mechanisms explaining the link do, however, exist. Besides the possible immediate fatal effect of an intracranial or torrential gastrointestinal or retroperitoneal haemorrhage, other consequences of bleeding may have specific detrimental effects in patients who have undergone a recent coronary event and revascularisation. Among these, the potential adverse effects of blood transfusions and the impact of the interruption of antithrombotic therapy are the most prominent.

Blood transfusion and clinical outcome

An analysis of over 24,000 patients from the GUSTO IIb, PURSUIT and PARAGON trials showed that patients with ACS who received at least one blood transfusion during their hospitalisation had a significantly higher unadjusted 30-day rate of death, myocardial infarction and death/myocardial infarction compared with patients who did not receive transfusions²⁷. The increased mortality associated with blood transfusion after PCI is to a large extent independent of the degree of the decrease in haematocrit and does, therefore, seem to be a direct negative effect of transfused blood²⁸. Impaired oxygen delivery by stored red blood cells has been repeatedly advocated as one of the mechanisms relating transfused blood to excess mortality. In fact, stored red blood cells are depleted of 2,3-diphosphoglycerate and, therefore, have a haemoglobin with higher affinity for oxygen and tend to release less oxygen to tissues. Other structural and biochemical changes occurring in stored red blood cells may also have a negative impact on the cells’ function in vivo, such as the loss of deformability or increased aggregability leading to enhanced plugging of the microvasculature, or the impaired capacity to deliver nitric oxide, an important defence mechanism against thrombosis^6,26. In favour of a detrimental role of transfused, stored red blood cells is the observation from a cohort of patients with a major diagnosis of cardiovascular disease admitted to acute care facilities who received blood transfusions during hospitalisation that the duration of red blood cell storage before transfusion was associated with the risk of in-hospital mortality²⁹. Based on these results, a prospective, large trial comparing the effectiveness of freshest available blood versus standard-issue blood on in-hospital mortality has been designed³⁰. Moreover, more strict rules on the management of bleeding and on the optimal transfusion strategy in patients with ACS have been advocated³¹.

Interruption of antiplatelet therapy and clinical outcome

It is now clear that patients with an ACS who have a bleeding while in hospital are significantly less likely to be discharged on an antiplatelet agent than patients without bleeding, and undertreatment persists for at least 6 months after discharge from hospital³². On the other hand the efficacy and, therefore, the benefit of antiplatelet therapy are especially great in the first weeks/months after an acute coronary event, i.e. in a condition associated with a high rate of cardiovascular events³³. Thus, inappropriate withdrawal of antiplatelet therapy in this critical period may generate a great burden of serious adverse cardiovascular events³⁴. An analysis of data from the PURSUIT, PARAGON-A, PARAGON-B and SYNERGY trials showed that the lack of antiplatelet therapy at hospital discharge is associated with a striking increase of long-term mortality; non-use of antiplatelet therapy at discharge was associated with worse outcomes, especially among patients treated with PCI³⁵.

Degree of platelet inhibition and the incidence of bleeding: the role of platelet monitoring

Over the last 10 years great interest has been raised by the finding of an association between insufficient platelet inhibition in patients on antiplatelet therapy, now defined as high on-treatment platelet reactivity, and a greater incidence of ischaemic cardiovascular events^34,36–38. In particular, non-responsiveness to clopidogrel has been consistently associated with long-term cardiovascular events³⁶, with a gradient showing that the level of residual on-treatment platelet reactivity is stepwise associated with long-term event rate³⁹. To a certain extent, non-responsiveness to clopidogrel is the consequence of a loss-of-function genetic polymorphism of liver enzymes involved in the in vivo metabolism of clopidogrel, a prodrug, into its active metabolite: several studies have shown a significantly higher incidence of ischaemic cardiovascular events in clopidogrel-treated patients carrying the CYPC19*2 loss-of-function mutation of the cytochrome p450 enzyme⁴⁰.

The association between excessive platelet inhibition detected by platelet function tests and the incidence of bleeding has been much less explored. Initial observations indicated that the risk of major bleeding in patients undergoing PCI and treated with aspirin plus a 600 mg loading dose of clopidogrel was significantly higher in patients with an enhanced response to clopidogrel, assessed by ADP-induced aggregation on a Multiplate analyser, than in the remaining patients⁴¹.

Similar data were later obtained by point-of-care platelet function testing with the Plateletworks device: the degree of inhibition of platelet activation correlated with postoperative chest drainage volume in patients treated with clopidogrel undergoing coronary artery bypass surgery⁴². In another, small study in patients undergoing PCI, post-treatment platelet reactivity correlated with the incidence of major bleeding, with high responders to clopidogrel experiencing a much higher rate of bleeding than middle or low responders⁴³.

Specular to the increase of cardiovascular events in carriers of a loss-of-function variant of the enzyme transforming clopidogrel into its active metabolite, a gain-of-function mutation, the CYP2C19*17 variant, has been associated with an increased rate of bleeding in clopidogrel-treated patients^44,45.

These data suggest that guidance of antiplatelet treatment based on platelet function testing might prove useful for avoiding bleeding events; however, prospective intervention studies are required to prove this hypothesis.

New and novel approaches to antiplatelet therapy associated with a potentially lower haemorrhagic risk

Given that the risk of bleeding appears to be indissolubly associated with current antiplatelet agents and based on the increasing awareness of the heavy burden that bleeding carries in patients with cardiovascular disease, great effort has been expended in the search for new pharmacological targets potentially enabling effective antithrombotic activity to be obtained with less bleeding^1,2,46 (Table I). Some of these new approaches have now been tested in large clinical trials while others are still in phase I or II studies in humans. It goes beyond the scope of the present overview to describe all the novel approaches to antiplatelet therapy, thus only two of these will be briefly discussed here: inhibition of the platelet thrombin receptor, protease-activated receptor-1 (PAR-1), and inhibition of the platelet GPIb/VWF interaction.

Table I.

Novel antiplatelet approaches undergoing clinical testing.

PAR-1 antagonists

Atopaxar

Vorapaxar

Inhibitors of the vWF-GPIb interaction

ARC1779 and ARC 15105 (aptamers)

ALX-0081 and ALX-0681 (anti A1 domain von Willebrand Factor nanobodies)

64B (humanised anti GPIb Fab)

GPG-290 (recombinant chimeric protein)

AJW200 (an IgG4 humanised monoclonal antibody anti A1 domain von Willebrand Factor )

82D6A3 (monoclonal antibody anti VWF A3 domain)

Inhibitors of the collagen/platelet interaction

Snake venoms (e.g. pallidipin, convulxin, jararhagin, NN-PF3, EMS16, rhodocetin)

Calin from Hirudo medicinalis

rLAPP from Haementeria officinalis

Aegyptin from Aedes aegypti

Monoclonal antibodies (e.g. JAQ1, OM2)

EXP2179 (active losartan metabolite)

DZ-697b (orally active inhibitor of collagen-platelet interaction)

Revacept (GPVI-Fc dimeric soluble form of GPVI)

PGE2 receptors (EP3) antagonists

DG-041

Nitric oxide donating agents

Nitroaspirin (NCX-4016)

Nitrostatins (NCX-6550; NCX-6560)

Protease-activated receptor-1 antagonists

Thrombin is the most powerful activator of platelets in humans and it acts by interacting with specific receptors on the platelet surface, among which protease-activated receptor-1 (PAR-1) seems to be the most relevant⁴⁷. A particular interest in this receptor derives from pathophysiological studies showing that this receptor is mainly involved in thrombus propagation and not in the formation of the initial haemostatic plug⁴⁸.

Based on these premises, small molecule, high affinity PAR-1 antagonists have been developed⁴⁹. Studies in animals and in healthy volunteers have shown that PAR-1 antagonist administration produces strong, long-lasting and selective inhibition of thrombin-induced platelet activation with no associated prolongation of the bleeding time⁴⁹. A phase II study with one of these molecules, vorapaxar, used on top of conventional antiplatelet and anticoagulant therapy in patients with ACS showed good tolerability, with no clear signs of increased bleeding and hints of enhanced antithrombotic efficacy⁵⁰. Two very large phase III clinical trials with vorapaxar have, therefore, been performed, one in patients with ACS (the TRACER trial) and one in patients with a previous myocardial infarction, stroke or with peripheral arterial disease on long-term secondary prevention (TRA-2P TIMI 50 trial)^51,52. Despite evidence of increased antithrombotic efficacy with vorapaxar in the TRA-2P trial, in both studies a significant increase of major bleeding was observed in vorapaxar-treated patients, and particularly worrisome was a striking increase of intracranial haemorrhage observed in both studies, especially in some categories of patients, such as those who had had a previous cerebrovascular event. A complete analysis of all the data from these two studies still needs to be completed, and it is possible that vorapaxar may prove advantageous in some specific subgroups of patients or in combination with specific antiplatelet drugs, but the expectations derived from preclinical and phase I–II studies in terms of low haemorrhagic risk do not appear to have been met.

Inhibitors of the GPIb/von Willebrand factor interaction

Targeting the interaction between VWF and its main platelet receptor, GPIb, is a potentially attractive approach⁵³. The interaction between VWF and GPIb is involved in the first phases of the formation of a thrombus on a damaged arterial surface and takes place especially in conditions of elevated shear stress, like those occurring in a stenosed atherosclerotic artery, and much less at intermediate or low shear stresses, such as those occurring in normal vessels, and in primary haemostasis. It is, therefore, conceivable that blocking the GPIb/VWF interaction will principally prevent arterial thrombosis without inducing bleeding. Several molecules, capable of interfering with the GPIb/VWF interaction either by blocking the A1 or A3 VWF domains or by interacting with platelet GPIb, such as aptamers, humanised monoclonal antibodies or nanobodies, have been developed (Figure 2)⁵³. All these agents can only be employed parenterally. Several preclinical studies, using different GPIb/VWF blockers, have given evidence of antithrombotic efficacy with little bleeding. Particularly interesting are data in ischaemic stroke models showing efficacy with no brain haemorrhage⁵⁴. Phase I studies in healthy volunteers and ongoing phase II studies in patients with ACS or with thrombotic thrombocytopenic purpura have so far confirmed safety. However, in preclinical models an increase of the doses of most of the GPIb/VWF antagonists has been associated with increased bleeding and it can be expected that the therapeutic window of these agents, i.e. the difference between the minimal dose preventing thrombosis and the first dose inducing bleeding, will not be particularly large. Whether this approach will produce enhanced antithrombotic efficacy with no increase of bleeding awaits confirmation from appropriate, large clinical trials.

Figure 2.

Strategies aimed at interacting with the binding of VWF to platelet GPIb/IX/V. Several agents blocking GPIb, the VWF-A1 domain or the VWF-A3 domain are shown.

Several other innovative approaches to antiplatelet therapy are being explored and some of them have the potential to be associated with less or no bleeding. Among these approaches, interference with the so-called primers of platelet activation, i.e. those mediators that accumulate in blood in pathological conditions which do not directly activate platelets but potentiate the platelet response to physiological stimuli, thus transforming a normal haemostatic response into unwanted thrombus formation^55–57, seems to be particularly appealing². Along this line, inhibitors of the PGE₂ receptors on platelets^2,58, antagonism of the effects of matrix metalloprotease-2 on platelets² and blockers of some signal transduction pathways in platelets^59,60, are examples of this approach. These innovative approaches are, however, all in the early stages of development.

If an antiplatelet drug with lower or no haemorrhagic risk emerges from this or other approaches, patients, especially those at high cardiovascular and haemorrhagic risk, would derive a great advantage in terms of net clinical benefit.

Concluding remarks

Antiplatelet therapy is the cornerstone of treatment for patients with conditions at high risk of ischaemic cardiovascular events, and its implementation and improved efficacy have contributed significantly to the decline of cardiovascular mortality observed in western societies over the last few decades⁶¹. Despite this great success, one important weakness of current antiplatelet regimens is their haemorrhagic risk and the serious consequences that may derive from bleeding. The way out of this problem may come from more precise identification of factors predicting bleeding in patients being treated with antiplatelet agents. Risk-scores for bleeding in patients with ACS have been developed⁶²: the validation of these in large prospective clinical studies, their utility in guiding antiplatelet treatment, the development of similar bleeding risk-scores for patients on long-term secondary prevention or for patients candidates to primary prevention and the optimisation of the management of bleeding complications, may all help to minimise the risks of antiplatelet therapy and to magnify its benefits.