Bleeding is the most important complication of oral anticoagulation (OAC) with vitamin K-antagonists. Whilst bleeding is unavoidably related to OAC, it may have a great impact on the prognosis of treated subjects by leading to discontinuation of treatment, permanent disability or death. The yearly incidence of bleeding during OAC is 2%-5% for major bleeding, 0.5%-1% for fatal bleeding, and 0.2%-0.4% for intracranial bleeding. While OAC interruption and/or antagonism, as well as administration of coagulation factors, represent the necessary measures for the management of bleeding, proper stratification of the individual risk of bleeding prior to start OAC is of paramount importance. Several factors, including advanced age, female gender, poor control and higher intensity of OAC, associated diseases and medications, as well as genetic factors, have been proven to be associated with an increased risk of bleeding. Most of these factors have been included in the development of bleeding prediction scores, which should now be used by clinicians when prescribing and monitoring OAC. Owing to the many limitations of OAC, including a narrow therapeutic window, cumbersome management, and wide inter- and intra-individual variability, novel oral anticoagulants, such as factor Xa inhibitors and direct thrombin inhibitors, have been recently developed. These agents can be given in fixed doses, have little interaction with foods and drugs, and do not require regular monitoring of anticoagulation. While the novel oral anticoagulants show promise for effective thromboprophylaxis in atrial fibrillation and venous thromboembolism, definitive data on their safety and efficacy are awaited.

The purpose of oral anticoagulation (OAC) therapy is to induce a controlled depression of blood coagulability in order to reduce the risk of thromboembolic complications in clinical conditions such as atrial fibrillation, mechanical heart valves, deep vein thrombosis and pulmonary embolism, and cardiogenic stroke. While newer oral anticoagulants, such as direct thrombin inhibitors (i.e., dabigatran) and direct factor Xa inhibitors (i.e., apixaban, rivaroxaban), have been recently evaluated in the clinical setting[1-5], vitamin K antagonists (VKAs), such as warfarin, acenocoumarol, and phenprocoumon, currently represent the standard drugs for OAC therapy.

The anticoagulant effect of VKAs is a consequence of their interference with the cyclic interconversion of vitamin K and its epoxide, by means of the inhibition of the vitamin K epoxide reductase enzyme (VKORC1), which in turn, is essential for the gamma-carboxylation of vitamin K-dependent coagulation factors, including factor II, VII, IX and X[6] (Figure 1). While acenocoumarol and phenprocoumon are preferred in some countries, warfarin is the VKA most commonly used in clinical practice[6]. Warfarin is a racemic mixture of two optically active isomers, the R and S enantiomers[6]. It is rapidly absorbed from the gastrointestinal tract, has high bioavailability, and reaches its maximal blood concentration about 90 min after oral administration[6]. Racemic warfarin has a half-life of 36-42 h, as compared to 8-24 h for acenocoumarol, and 80-270 h for phenprocoumon, circulates bound to plasma proteins, and accumulates in the liver, where the two enantiomers are metabolized by different pathways[6]. The S enantiomer, which is about three times more potent than the R enantiomer, is primarily metabolized by the CYP2C9 enzyme of the cytochrome P450 system, whereas the R enantiomer is metabolized by the CYP1A2 and CYP3A4 enzymes[6].

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Mechanism of action and metabolism of warfarin. VKORC1: Vitamin K epoxide reductase enzyme.

In order to maximize protection against thromboembolic complications while minimizing the risk of bleeding associated with VKA therapy, the intensity of OAC should be maintained within a narrow therapeutic range (TTR). Apart from non-bileaflet mechanical heart valves and mechanical heart valves in the mitral position, where a higher intensity of OAC therapy is required, an international normalized ratio (INR) of 2.0-3.0 has long been identified as the optimal TTR for most clinical conditions at risk for thromboembolic events[6]. Nonetheless, bleeding represents the major complication of OAC therapy, even when OAC is properly prescribed[7].

The reported incidence of bleeding during OAC therapy with VKAs is highly variable in published studies. This variability may be accounted for by the differences in the definition and classification of bleeding (Table 1). More importantly however, the differences in the reported hemorrhagic rates are more likely to be attributed to differences in study design and patient population. In randomized clinical trials, where highly selected patients are enrolled, rates of bleeding are expected to be lower than in observational studies, where patients commonly encountered in clinical practice, including those with individual risk factors for bleeding, are included. As an example, in six pivotal trials evaluating the effect of warfarin compared to placebo in patients with atrial fibrillation, only 12.6% out of the 28 787 patients screened were finally included[8]. Whether or not monitoring of OAC therapy is carried out in specialized services, where a better and more stable control of the INR is generally achieved, may also have an impact on the occurrence of bleeding complications[9].

The expected incidence of bleeding during long-term VKA therapy is about 10%-17% per year for all events, 2%-5% per year for major bleeding, and 0.5%-1% per year for fatal bleeding[10]. The reported occurrence of intracranial hemorrhage (ICH), which represents the mostly feared bleeding complication of VKA therapy because of its high disability and/or fatality rate, is in the range of 0.2%-0.4% per year[7,11].

The occurrence of bleeding during OAC therapy with VKAs has relevant prognostic and management implications. Major bleeding can be life-threatening, such as when occurring at critical sites like the head, the pericardium or the pleura, or when leading to the development of hemorrhagic shock. Also, major bleeding complications are generally associated with VKA discontinuation, which in turn, contributes to adverse outcomes by leaving the patient exposed to an increased risk of thromboembolism. In a recent meta-analysis of clinical trials reporting on OAC therapy with VKAs for venous thromboembolism, a case-fatality rate of major bleeding of 13.4% (95% CI: 9.4%-17.4%) has been reported in all patients, with a rate of ICH of 1.15% per year (95% CI: 2.5%-21.7%)[12]. Regardless of the severity however, hemorrhagic complications of OAC therapy with VKAs may be important for the related inconvenience to the patients. The perceived higher risk of bleeding of VKA therapy limits its more widespread use, therefore excluding many patients from the benefits of such therapy. It has been shown that the occurrence of VKA-associated adverse events has an impact on the prescribing of this treatment, since physicians are less likely to prescribe VKAs after observing a bleeding complication during VKA treatment in their patients[13].

Management of bleeding occurring during OAC therapy with VKAs generally consists of discontinuation and/or antagonism of VKAs, as well as of local measures (i.e. endoscopic treatment and/or surgical hemostasis) and proper transfusion procedures. Discontinuation of VKAs allows for subsequent normalization of INR, which may require several hours owing to the prolonged half-life of VKAs. According to available data, temporary interruption of VKA therapy because of major bleeding or trauma appears not to be associated with an increased risk of subsequent thromboembolic complications[14,15]. In a population of 28 patients with prosthetic heart valves receiving warfarin and hospitalized for major hemorrhage who were retrospectively evaluated, discontinuation of OAC therapy for 1 d to > 3 wk (mean duration 15 ± 4 d) was not associated with thromboembolic complications at 6-mo follow-up[14]. In a retrospective, population-based, cohort study evaluating 8450 patients aged > 65 years on warfarin for various reasons and surviving a major trauma, the incidence of stroke at a mean follow-up of 3.3 years did not differ between patients who discontinued warfarin as compared to those who continued warfarin [hazard ratio (HR), 0.99; 95% CI: 0.82-1.21], while the incidence of major hemorrhages was significantly lower (HR, 0.69; 95% CI: 0.54-0.88) and that of venous thromboembolism was significantly higher (HR, 1.59; 95% CI: 1.07-2.36)[15]. Antagonism of the VKA anticoagulation effect is obtained by directly administering vitamin K[16]. Vitamin K can be given orally or intravenously, with the parenteral route having the advantage of a more rapid onset of action[17]. After intravenous administration of vitamin K, the INR will start to drop within 2 h and will be normalized within 12-16 h[18], whereas after oral administration it will take up to 24 h to normalize the INR[19]. While low (1-2.5 mg) to moderate (2.5-5 mg) doses of vitamin K given orally are indicated for the management of non-emergency bleeding, slow intravenous infusion of 10 mg vitamin K should be given in emergency situations[6]. Higher doses of vitamin K are equally effective but may induce resistance to VKAs for more than 1 wk[20], and are therefore not recommended. In emergency situations, immediate correction of high INR should be pursued also by the administration of vitamin K-dependent coagulation factors. These factors are present in fresh frozen plasma, which however is inconvenient to use owing to the very large amount needed, the prolonged time required for administration, and the associated risk of fluid overload[21]. Therefore, prothrombin complex concentrates, most of which contain all vitamin K-dependent coagulation factors, are more useful and should be administered either according to fixed dose schemes or, preferably, by individualized dosing regimens based on INR value and body weight[22].

Numerous factors, including individual characteristics, intensity, timing and quality of OAC therapy, and use of concomitant medications, have been established as impacting on the risk of bleeding during VKA treatment (Table 2).

Stratification of the individual bleeding risk prior to initiation of OAC therapy with VKAs is another important measure to adopt for prevention of hemorrhagic complications[53]. To assist clinicians in this task, several schemes have been developed (Table 3).

The modified outpatient bleeding risk index (mOBRI) has been prospectively derived and extensively validated in patients with different indications for VKAs[54]. Also, the mOBRI was evaluated in a “real-life” setting of anticoagulation monitoring (i.e., primary care physicians or pharmacist-run anticoagulation clinic) and the assessment of bleeding was blinded[54]. The hepatic or renal disease, ethanol abuse, malignancy, older, reduced platelet count/function, re-bleeding, uncontrolled hypertension, anemia, genetic factors, excessive fall risk, and stroke score has been derived from a retrospective chart review of nationwide registry data of patients with atrial fibrillation[55]. The retrospective design, the inclusion of genetic factors (which are rarely investigated in common practice), and of anemia and platelet data (also infrequently available), limit the usefulness of this score, as well as the lack of inclusion of important factors such as the concomitant use of antiplatelet agents and/or other drugs. The schema by Shireman et al[56] is also derived from a retrospective evaluation of nationwide registry data, and incorporates eight risk factors for bleeding. As regards limitations, the follow-up period was relatively short (i.e. 90 d), the outcome assessment was not blinded, and the quality of anticoagulation control and the use of comedications were not reported[56]. Finally, the individual determination of this score requires a complex mathematical calculation, which makes it rather impractical[56]. The hypertension, abnormal liver or renal function, stroke, bleeding, labile INRs, elderly, drugs or alcohol (HAS-BLED) score has been recently developed for patients with atrial fibrillation, based on the prospective Euro Heart Survey on atrial fibrillation[57]. Compared with the previous three scores, the HAS-BLED score was recently demonstrated to have better predictive ability for bleeding among all patients enrolled in the SPORTIF III and V trials, as well as among warfarin-naïve patients and patients treated with warfarin and aspirin[58]. The HAS-BLED score appears easy and user-friendly for everyday application, owing to the small number of variables to memorize and the common availability of all the information required for risk calculation. Finally, the HAS-BLED score allows for periodical re-assessment of a patient’s bleeding risk as it also takes into account the quality of INR control.

When applying the different bleeding scores, a wide variability in the prevalence of patients determined to be at low-, moderate-, and high-risk is observed because of differences in the risk factors comprising each schema, the various definitions of major bleeding, the different lengths of follow-up in the cohorts, and the different populations evaluated in the studies. Furthermore, no data on the impact of these scores on patient outcomes are currently available, so that none of the proposed bleeding risk scores can currently be recommended for widespread clinical use. Nonetheless, it must be emphasized that bleeding risk stratification should represent an integral part of the management of OAC therapy, and application of these scores may assist clinicians in their decision-making in everyday clinical practice.

While VKAs are the standard drugs for OAC therapy, they have several major limitations, including the narrow therapeutic window, the cumbersome management, and the wide inter- and intra-individual variability. Therefore, more effective and/or safer and/or easier to use oral anticoagulants have long been sought. Recently, two main classes of new non-VKA oral anticoagulants (i.e. factor Xa-inhibitors and direct thrombin inhibitors) have been developed. These agents can be given in fixed doses, have little interaction with foods and drugs, and do not require regular monitoring of anticoagulation[59].

Apixaban, rivaroxaban and edoxaban are orally active, direct inhibitors of factor Xa. While being currently evaluated in several large, double-blind trials in patients with atrial fibrillation or venous thromboembolism, some evidence about the efficacy and safety of these drugs has already been reported. In the EINSTEIN-DVT and EINSTEIN-Extension studies[1], 3449 patients with acute deep vein thrombosis received, in an open-label manner, rivaroxaban or standard treatment (i.e. initial course of enoxaparin followed by VKAs) for 3, 6 or 12 mo. The 602 patients who had completed 6 to 12 mo of treatment then entered a double-blind study with randomization to rivaroxaban or placebo for an additional 6 or 12 mo. In the whole population, the efficacy of rivaroxaban on recurrent venous thromboembolism was non-inferior to the enoxaparin/VKA regimen (2.1% vs 3.0%, P < 0.001). The safety, as expressed by the on-treatment occurrence of major bleeding, was identical (8.1% in both groups). In the extended-treatment group, rivaroxaban was significantly more effective (1.3% vs 7.1%, P < 0.001), however, there was an increased, albeit non significant, incidence of major bleeding (0.7% vs 0%). In the AVERROES trial[2] 5599 patients with atrial fibrillation, who were unsuitable for or who were not willing to receive VKAs were randomized to apixaban or aspirin. At the 1.1-year follow-up, the treatment with apixaban was significantly more effective on the incidence of stroke/systemic embolism (1.6% vs 3.7%, P < 0.001). No differences in the rates of major bleeding were observed between the apixaban and aspirin groups (1.4% vs 1.2%, P = 0.57). In the double-blind ROCKET-AF trial[3], which enrolled > 14 000 atrial fibrillation patients, and investigated the efficacy (incidence of stroke/systemic embolism) and safety (incidence of bleeding) of rivaroxaban compared to warfarin, rivaroxaban was more effective (1.71% vs 2.16%, P < 0.001 for non-inferiority), without no significant difference in the rate of major bleeding (3.60% vs 3.45%).

Dabigatran is an orally active, direct thrombin inhibitor, which has been tested for the long-term treatment of patients with nonvalvular atrial fibrillation and venous thromboembolism. In the RE-LY trial, 18 113 patients with atrial fibrillation were randomized to two blinded doses of dabigatran against open-label warfarin[4]. At 2-year follow up, the primary outcome of stroke and systemic embolism was lower both with low-dose and high-dose dabigatran compared to warfarin (1.53% and 1.11% vs 1.69%, P < 0.001 for non-inferiority and P < 0.001 for superiority, respectively). The incidence of major bleeding was lower in the low-dose group (2.71% vs 3.36%, P = 0.03) and similar to warfarin in the high-dose group (3.11% vs 3.36%, P = 0.31). Among the 2564 patients with acute venous thromboembolism enrolled in the double-blind RE-COVER study, dabigatran was as effective as warfarin for secondary prevention of recurrent venous thromboembolism at 6-mo follow-up (2.4% vs 2.1%). The incidence of major bleeding was also comparable in the two groups (1.6% vs 1.9%)[5].

In the light of the promising outcomes, as well as of the superior ease of use, the novel non-VKA oral anticoagulants represent an attractive alternative to VKAs for OAC therapy. Nonetheless, additional data on the impact of the lack of a method to monitor the intensity of anticoagulation, lack of a specific antidote, and little information regarding the risk/benefit ratio of their combination with more potent antiplatelet agents, such as the increasingly used prasugrel or ticagrelor, are necessarily awaited before these drugs can be truly considered substitutes for VKAs.

OAC therapy with VKAs is the most effective available treatment for prevention of thromboembolic complications in frequent clinical conditions, such as atrial fibrillation, venous thromboembolism, prosthetic heart valves, and cardiogenic stroke. Such efficacy comes at the price of an increased risk of overall, major and ICH bleeding. While novel, non-VKA oral anticoagulants, such as direct thrombin inhibitors and direct factor Xa inhibitors, have been showing promise in the setting of atrial fibrillation and venous thromboembolism, VKA treatment should currently be offered to all patients at high risk of arterial/venous thromboembolism because of the established benefit outweighing the risk. However, meticulous care should be directed towards accurate bleeding risk stratification and minimization and/or control of all the numerous factors known to influence the risk of hemorrhagic complications.