Reviewing the reality: why we need to change

For nearly 60 years, vitamin K antagonists (VKAs) such as warfarin have been the mainstay of oral anticoagulation in a range of thromboembolic disorders, ranging from stroke prevention in atrial fibrillation (AF) to the treatment and secondary prevention of venous thromboembolism. Although highly effective, warfarin has numerous limitations that have resulted in its underuse and the imposition of considerable burdens on both patients and healthcare systems. In the case of stroke prevention in AF, the argument has been made that warfarin use can be maximized and its limitations minimized by using several strategies, such as improving international normalized ratio (INR) control through greater use of anticoagulation clinics or patient self-management, and greater adherence to relevant treatment guidelines.1 But is this realistic? Can the limitations of VKAs such as warfarin really be effectively and properly minimized in everyday clinical practice? In this article, the author provides a counter-argument that warfarin is difficult to use in clinical practice, that its limitations make it difficult to manage, and that we need new oral agents with improved benefit–risk profiles and increased consistency and predictability when compared with warfarin. To understand the need for new agents, one must first understand how warfarin works. To this end, it is necessary to review the history of warfarin, and its pharmacodynamic and pharmacokinetic profiles, so that it becomes clear that the concerns and challenges that face physicians and patients are very real and are a direct result of the characteristics of warfarin. More than 80 years ago, livestock farmers in North Dakota, United States, observed severe, unexplained bleeding in cattle, which was later linked to their diet of fermented hay or silage made from sweet clover, grown as a substitute for corn at this time.2 In 1939, the coumarin dicoumarol was identified as the active agent in sweet clover responsible for the bleeding disorder.3 Several coumarin derivatives have since been synthesized, the most potent being warfarin (derived from the acronym for Wisconsin Alumni Research Foundation, WARF).4 It was initially used as a rat and mouse poison, but the survival of a man after an attempted suicide by the use of a large amount of warfarin-based rodenticide led to clinical trials of warfarin as an anticoagulant in humans.5 To understand the limitations of warfarin, we need first to appreciate the coagulation system and the sites of action of warfarin. The coagulation system is designed to prevent bleeding at the site of vessel injury through the formation of a blood clot. Clot formation occurs as a result of interactions between three components: the blood vessel wall, cellular components within the blood (predominantly platelets in the arterial circulation and red cells in the venous circulation), and plasma clotting factors. A detailed review of the blood clotting process is beyond the scope of the present article, and readers are referred to some reviews for further information.6–9 Focusing briefly on the plasma clotting factors, these factors are involved in a proteolytic ‘cascade’ activation pathway that ultimately results in the production of thrombin, via the intrinsic and extrinsic pathways (Figure 1). Thrombin then converts soluble fibrinogen into insoluble fibrin, forming a clot. Under normal haemostatic conditions, excessive clotting is prevented by several mechanisms, including the following: antithrombin III, which inhibits factors IX, X, XI, and XII and thrombin; tissue factor pathway inhibitor, which inhibits active factor X; the protein C, protein S, and thrombomodulin complex, which inhibits factors VIII and V; and of course the fibrinolytic system. However, in a condition such as AF, the haemostatic balance is disrupted as a result of stasis within the fibrillating left atrial appendix, leading to inappropriate clot formation.10,11 Effectively, the balance has been tipped in favour of forming blood clots. Hence, our role is to restore that balance by using anticoagulants. What anticoagulant strategies are available to prevent clot formation and to restore the haemostatic balance? One approach has been the use of heparins (unfractionated and low-molecular-weight heparins), which indirectly act to enhance the activity of antithrombin III.12 However, the need for parenteral administration has limited their use, particularly in the chronic/outpatient setting.13,14 Another strategy uses warfarin and other VKAs. Warfarin exerts its anticoagulant effect by inhibition of the synthesis of several vitamin K-dependent proteins—the coagulation factors VII, IX, and X, thrombin (factor II) and the anticoagulation proteins C and S.15 The mechanism of action of warfarin and how this leads to several of the complications seen with this agent is discussed subsequently. Vitamin K-dependent proteins require γ-carboxylation to be fully active, and this process is coupled with oxidization of vitamin K (Figure 2). Oxidized vitamin K cannot be used again; therefore, the body recycles it by reducing the vitamin K to its natural state. VKAs prevent the recycling of vitamin K by blocking this conversion of oxidized vitamin K back to the reduced form. This in turn blocks γ-carboxylation of vitamin K-dependent proteins, resulting in the synthesis of proteins that have reduced activity.16 This mechanism of action means that warfarin has a delayed onset of action, as the complete effect of warfarin is only achieved after clearance of currently active vitamin K-dependent coagulation factors has taken place. It can take several days for the maximal effect of warfarin to be achieved owing to the long half-lives of some vitamin K-dependent coagulation factors (Table 1).The half-lives vary for the different coagulation factors, e.g. factor VII has the shortest half-life (4–6 h), and hence it will be depleted first, but others have longer half-lives (up to 72 h). Therefore, the full effect is not achieved for several days. Another complicating factor is that proteins C and S, which are also inhibited by warfarin, are endogenous anticoagulants that reduce thrombin generation. Protein C has a short half-life (only 8 h), which means that warfarin therapy may produce a hypercoagulable state initially until more coagulation factors are metabolized and removed from the system.16 Owing to this delayed onset of action and also the potential hypercoagulable state, overlap with heparin is required, typically for 3–5 days, to protect patients until the full effect of warfarin is realized.16 Once an anticoagulant effect is established, the main complication of warfarin therapy is bleeding, with the bleeding risk increasing substantially with the intensity of anticoagulation.16 A narrow optimal INR range of 2.0–3.0 has been identified as providing effective anticoagulation at the lowest risk of bleeding.1,16 Patients therefore require regular coagulation monitoring and frequent dose adjustments to get the level of anticoagulation to within the therapeutic range. Bleeding risk is particularly increased in patients receiving concomitant antiplatelet drugs,17 those aged >65 or those with prior stroke, renal impairment, or anaemia.18,19 A major drawback to the VKA mode of action is that subjects receiving long-term therapy will be sensitive to fluctuating dietary vitamin K. Vitamin K in the diet is derived predominantly from phylloquinones in plant material,20 which are found in a variety of foods including cauliflower, green cabbage, seaweed, broccoli, and spinach.21 Dietary vitamin K is converted in the liver to a reduced form by a warfarin-insensitive system that allows the synthesis of fully active vitamin K-dependent coagulation factors, thereby counteracting the anticoagulant effect of warfarin. This can lead to intra-individual inconsistency in the dose–response relationship of warfarin. Therefore, variable intakes of dietary vitamin K can create significant fluctuations in the anticoagulation effects despite the same dose of warfarin. Warfarin is a racaemic mixture of two enantiomers, R- and S-warfarin, the S-form being three to five times more potent than the R-form.22,23 The two forms are present in roughly equal proportions, have half-lives of 36–42 h, circulate bound to plasma albumin, and ultimately accumulate in the liver where they are metabolized.16 Warfarin is differentially metabolized by a variety of cytochrome P450 (CYP) enzymes, depending on the enantiomer being metabolized (stereoselective metabolism) and the site of metabolism (regioselective metabolism) (Figure 3).24In vitro studies have shown that S-warfarin is metabolized predominantly by CYP2C9 and to a lesser extent through CYP3A4.24R-warfarin, on the other hand, is metabolized predominantly through CYP1A2 but also through CYP3A4 and CYP2C19. Therefore, metabolism of warfarin involves multiple cytochromes and, as a consequence, is particularly prone to pharmacokinetic drug interactions. The drugs most likely to interact are those that are also metabolized by the CYP system. Some of these drugs are CYP inhibitors, resulting in decreased warfarin metabolism and therefore higher serum concentrations of warfarin. Inhibition of S-warfarin metabolism is more important clinically, owing to the increased pharmacological potency compared with the R-form. Other drugs induce (accelerate) the CYP system, causing warfarin to pass through the body faster, resulting in lower serum levels of warfarin. With all these multiple points of interaction, drugs from many different classes can alter warfarin metabolism. For example, phenylbutazone (a non-steroidal anti-inflammatory drug), sulfinpyrazone (used for the treatment of hyperuricaemia), and the antibiotics metronidazole and trimethoprim each potentiate the anticoagulant effect of warfarin through inhibition of S-warfarin metabolism.16,25–27 In contrast, drugs such as the barbiturate depressants, the antibiotic rifampicin, and the anticonvulsant carbamazepine inhibit the anticoagulant effect of warfarin through increased hepatic clearance.16 The list of potential drug interactions is extensive;28 in fact, warfarin interacts with almost every class of drug known. This is an important consideration for treating conditions such as stroke prophylaxis in AF, where patients at the highest risk tend to be elderly and are more likely to have co-morbidities requiring polypharmacy. Warfarin therapy is further complicated by genetic and environmental factors. Genetic polymorphisms resulting from single amino acid substitutions can reduce the metabolic activity of CYP2C9, leading to an increased level of S-warfarin and an increased frequency of bleeding.29 Wild-type CYP2C9 (CYP2C9*1) has at least five allelic variants (CYP2C9*2–CYP2C9*6) that underlie inter-individual and ethnic differences. For example, CYP2C9*2 and CYP2C9*3 are significant among white populations, CYP2C9*4 has been exclusively identified in Japanese individuals, and CYP2C9*5 and CYP2C9*6 are found among African-Americans.30 Carriers of such variant alleles are at particular risk of adverse events when a drug has a narrow separation of antithrombotic and haemorrhagic effects, as is the case with warfarin. This was demonstrated by a case of a 90-year-old patient treated with 5 mg of warfarin per day, who was hospitalized for gastro-intestinal bleeding with an INR of 66.31 Warfarin could still be detected in this individual 11 days after the last dose, suggesting reduced clearance of the drug. Genetic analysis revealed that this patient was a carrier of the two variant alleles CYP2C9*2 and CYP2C9*3.31 There is also reduced warfarin clearance with age, which is reflected in exaggerated anticoagulant responses, a further complication for warfarin use in the elderly.32,33 Hereditary resistance to warfarin has also been demonstrated,34–36 where patients require much larger doses (five- to 20-fold higher) of warfarin to achieve an anticoagulant effect.16 This warfarin resistance is attributed to a reduced affinity of warfarin for its hepatic receptor. Cigarette smoking can induce CYP1A2 activity, the major cytochrome responsible for R-warfarin metabolism. With increased smoking, the CYP1A2 increases its metabolism of R-warfarin, which translates into a need for higher dosages. During longer non-smoking periods, the enzyme activity slows down and thus less warfarin is needed. Therefore, a change in smoking habit may affect warfarin coagulation response. After smoking cessation, for example, patients need to be carefully monitored and warfarin doses reduced accordingly.37 A detailed look at the characteristics of warfarin and its pharmacological profile has highlighted its unpredictable anticoagulant response and the underlying reasons for it. The close coagulation monitoring required to maintain the level of anticoagulation within the narrow therapeutic range can cause inconvenience to the patient and required the infrastructure to deal efficiently with such intense patient management. In fact, even with weekly monitoring, it is still possible for the INR to be out of range because of changes in diet or in intermittent use of other medications that affect warfarin levels. Therefore, even more frequent testing may not eliminate the difficulties that warfarin poses. The limitations associated with warfarin therapy have resulted in the global problem of under-treatment with anticoagulants for the prevention of strokes in AF. In Italy, for example, only about one in five hospitalized patients with AF will receive oral anticoagulation and only 25% of chronic AF patients receive a VKA.38 The United States has the best rate of VKA use in AF, but this still only amounts to ∼40% of patients with AF receiving oral anticoagulation.39 Physician surveys suggest that fear of bleeding complications is the major factor responsible for under-prescription.40,41 Given that warfarin therapy is proved to be effective and safe if maintained within an INR range of 2.0–3.0, is this fear of bleeding complications unwarranted? The unpredictable nature of warfarin means it is difficult to monitor and the therapeutic range is difficult to achieve, even in controlled trial settings. For example, the SPORTIF II trial was a small study involving approximately 50–100 warfarin-treated patients with AF.42 The study protocol and controlled trial conditions meant that experienced nursing staff carried out regular monitoring and dose adjustments, in order to maintain an INR within the target range. Since the trial involves a relatively small number of individuals, control of INR would be expected to be better than that generally achieved in clinical practice. Despite this, an INR of 2.0–3.0 was only achieved 44% of the time (Figure 4). Patients received sub-therapeutic levels 38% of the time, and 18% of the time the INR was above 3.0, increasing the risk of major haemorrhage. This difficulty in achieving INR is reflected in the frequency of major bleeding observed with warfarin at the start of outpatient therapy. Follow-up information for 562 patients initiating warfarin therapy revealed a cumulative incidence of 11% for major bleeding in the first 12 months.19 Furthermore, a recent prospective analysis of more than 18 820 patients admitted to two hospitals in the UK over a 6-month period found that in 80% of cases hospitalization was due to adverse drug reactions43 and that warfarin alone was responsible for 10.5% of cases, with adverse reactions including gastro-intestinal bleeding and haematoma. Despite the proven benefit of warfarin, this level of morbidity and the associated costs explains the reluctance of physicians to start therapy and highlights the need for drugs with improved benefit–risk ratios. Warfarin therapy has been invaluable as an oral anticoagulant over the past 60 years; however, its limitations have provided the stimulus for the development of new oral anticoagulants that will overcome the drawbacks of VKA therapy. These agents are being developed to have limited or no dietary interactions and a low potential for drug interactions, with a predictable pharmacokinetic profile. The consistent anticoagulant response would allow fixed dosing without coagulation monitoring. Furthermore, a fast onset of action and oral administration would allow utility in both acute and chronic settings. As described earlier, the coagulation cascade ultimately leads to thrombin generation, which in turn leads to platelet activation and fibrin formation. The key role of thrombin in the coagulation process meant that it was identified as a promising therapeutic target and led to the development of the direct thrombin inhibitor class of anticoagulant. The first oral agent in this class, ximelagatran, represents the first-ever alternative to VKAs for oral anticoagulation. Ximelagatran was developed to overcome many of the limitations of warfarin and other VKAs and has the potential to maximize anticoagulation use in patients with AF, including better adherence to established treatment guidelines. Figure 1 Disruption of the coagulation cascade by warfarin. Figure 2 The vitamin K cycle and its link to carboxylation of coagulation factors. Vit, vitamin. Reproduced with permission from Hirsh J. et al.44 American Heart Association/American College of Cardiology Foundation guide to Warfarin therapy. Circulation 2003;107:1692–1711. Figure 3 R- and S-warfarin and the major CYP enzymes responsible for their metabolism. Figure 4 The SPORTIF II trial: proportion of patients within target INR range. Reprinted from J Am Coll Cardiol, Vol 41, Peterson P. et al. Ximelagatran versus warfarin for stroke prevention in patients with nonvalvular atrial fibrillation. SPORTIF II: A dose guiding tolerability and safety study, pp. 1445–1451. © (2003), with permission from the American College of Cardiology foundation. The variable half-lives of vitamin K-dependent factors The variable half-lives of vitamin K-dependent factors