Atrial fibrillation (AF) is the most frequent sustained arrhythmia. The estimation is that five million people suffer the disease in Europe,1 although this number may be even higher because many AF episodes are silent but still associated with a risk of complications.2–4 AF prevalence is increasing in developed countries in part due to the progressive increase in the mean age of the western population but also due to unknown factors. This progressive increase and the negative impact of this arrhythmia on the prognosis of patients makes AF, together with heart failure and diabetes, one of the main healthcare problems for the coming decades.5,6 This has led to intense research into the main aspects of this rhythm disorder, one of them being thromboembolism prevention.6 This article reviews thromboembolism in AF and the new drugs for its prevention.
Atrial Fibrillation and Thromboembolism
Thrombus formation within the vascular system is dependent on three main factors: blood stasis, cardiovascular wall damage and coagulation status. Blood stasis is always present in patients with AF but atrial wall damage (post-cardioversion transient mechanical dysfunction, ischaemic or rheumatic lesions, etc.) and markers of increased coagulability are also commonly found.6–10 For these reasons, atrial thrombus formation is common in patients with AF (see Figure 1) and represents the main cause of stroke. However, other sources of thrombosis and systemic embolism, such as cerebral arteriosclerosis, carotid or ascending aorta ulceration, left ventricle thrombosis or venous system embolism in patients with patent foramen ovale also have to be considered. AF patients have a four to five times higher risk of stroke than the general population, even if AF is silent,11,12 and one out of six patients with stroke has AF.6
Thromboembolic Risk Factors in Atrial Fibrillation
Several factors increase thromboembolic risk in patients with AF. Rheumatic valve disease (mitral stenosis), mechanical prosthetic heart valve, and history of past embolic events are the most relevant and are considered sufficient by themselves to indicate permanent anticoagulation. Other clinical thromboembolic risk factors are older age, hypertension, congestive heart failure, diabetes, coagulation disorders, difficult management of anticoagulation with vitamin K antagonist and recent changes in anticoagulation treatment.
Treatment of some of these risk factors, such as mitral valve stenosis or arterial hypertension, is per se effective at reducing the risk of embolism. Other defects detected by echocardiography have been also associated with thromboembolism in patients with AF.13–15 Left atrium thrombi, spontaneous echocontrast in the left atrium, left atrium dilatation and inter-atrial septum aneurysms are the most relevant. Of these, left atrium thrombus carries the highest embolic risk but can be hardly identified by echocardiography without transesophageal echocardiography.
Congestive Heart Failure, Hypertension, Age Greater than 75, Diabetes, and Prior Stroke or Transient Ischemic Attack Score
All the risk factors mentioned are associated with systemic embolism and risk increase when they come together. However, the use of risk scores is more convenient because they can be easily used in most patients, facilitate estimation of individual risk, allow comparison of different trials and identify the best candidates for anticoagulation. The most used is the Congestive Heart Failure, Hypertension, Age Greater than 75, Diabetes, and Prior Stroke or Transient Ischemic Attack (CHADS2) score, which evaluates the presence of congestive heart failure, hypertension, age, diabetes, and prior stroke or cerebral transient ischaemic accident.16–20 All of these latter factors are easily identified in all patients and their combination allows more precise risk estimation than other scores. Each positive factor adds one point to the score except for stroke, which counts for double and is annotated with a subscript 2 in the acronym CHADS2. Table 1 shows how stroke risk increases with CHADS2 score in patients taking and not taking aspirin. From the practical point of view, three groups of CHADS2 patients are considered: low risk (score 0), medium risk (score 1–2) and high risk (score >2). In addition to this, patients with mechanical prosthetic heart valve, mitral valve stenosis or left atrium thrombi have to be considered also high risk irrespective of their CHADS2 score.6
Antithrombotic Treatment for Embolism Prevention
Antithrombotic drugs are the mainstream therapy for embolic prevention in patients with AF. Different trials have tested drugs such as aspirin, trifusal, clopidogrel and different anticoagulants during the last 50 years. Interestingly, until the 1990s antithrombotic drugs were mostly used in high-risk patients (prosthetic heart valve, mitral stenosis or prior embolism) but most patients with non-valvular AF received no treatment with any of these drugs.
However, clinical practice has changed dramatically in the last 20 years due to the results of several clinical trials that demonstrated the benefit of both antiplatelet and anticoagulant drugs in patients with non-valvular AF.6 The clinical use of these drugs is based on the risk–benefit ratio, where benefit is the reduction of stroke and other systemic embolic events and risk is mostly driven by the increase in bleeding events. In general, antiplatelets are indicated for low-risk patients, with anticoagulants being the drug of choice for moderate- and high-risk patients.
Age, hypertension, diabetes, renal failure, low bodyweight, female gender, bleeding history, poor anticoagulation control, concomitant use of several antithrombotic drugs, instability and poor muscular co-ordination increasing the risk of trauma, and high levels of anticoagulation are the main factors associated with severe bleeding.6,20–23 Different risk scores are available for clinical practice but never replace clinical judgment in the calculation of risk–benefit in each individual patient.6 In addition, it should be noted that patients with a moderate or high risk of bleeding were excluded from all clinical trials showing a benefit of antithrombotic drugs.
Benefits of Aspirin
Embolic risk reduction with aspirin is modest in AF patients. A meta-analysis of the main randomised trials comparing aspirin versus placebo for thromboembolism prevention in patients with AF24 showed slight reduction of stroke, either ischaemic or haemorrhagic, although risk reduction is lower than that observed with warfarin.24
Benefits of Vitamin K Antagonists
A meta-analysis of vitamin K antagonists versus placebo showed the high efficacy of the former to prevent thromboembolism in patients with AF.24 These studies made anticoagulation the standard therapy for moderate- and high-risk patients with AF. The level of anticoagulation has been evaluated in several trials, and 2.5 is considered to be the best score, with a range between two and three considered to be optimal.6
A recent study evaluated the CHADS2 score in a population of 13,559 patients with non-valvular AF and 66,000 persons/year follow-up. Net clinical benefit was defined as the annual rate of systemic embolism, stroke or cerebral bleeding, the latter weighted 1.5 due to the higher clinical relevance compared with stroke. Net clinical benefit for warfarin was 32% with a positive correlation with the CHADS2 score and with similar results when cerebral bleeding was weighted one of two (see Figure 2).25 Net clinical benefit was also different in each group of patients.
Other Antithrombotic Drugs
The efficacy of several antiplatelets and anticoagulants different to aspirin and warfarin was compared against placebo in several trials. The Studio Italiano Fibrillazione Atriale (SIFA) trial26 compared indobufen, a cyclooxygenase inhibitor different to aspirin. Indobufen showed a similar haemorrhagic complication rate to warfarin but the number of embolic events was inferior with the latter.
The combined effect of clopidogrel and aspirin was compared with warfarin in the Atrial Fibrillation Clopidogrel Trial with Irbesartan for Prevention of Vascular Events, substudy W (ACTIVE-W).27 The combination of the two antiplatelets was less efficacious than warfarin and the safety committee of the trial had to end it prematurely.
The Atrial Fibrillation Clopidogrel Trial with Irbesartan for Prevention of Vascular Events, substudy A (ACTIVE-A) studied the effect of combining clopidogrel or placebo with aspirin in 7,554 patients who had contraindications to warfarin.28 Clopidogrel plus aspirin showed a lower stroke rate than placebo plus aspirin (2.4 versus 3.3% per year; relative risk [RR] 0.72; 95% confidence interval [CI] 0.62–0.3; p<0.001) but the incidence of severe haemorrhages increased (2 versus 1.5% per year; p<0.001). The interpretation of the two ACTIVE trials shows that the combination of clopidogrel and aspirin is better than aspirin alone but only in patients who are poor candidates for warfarin, since this is superior to the former combination of antiplatelet drugs.
The National Study for Prevention of Embolism in Atrial Fibrillation (NASPEAF)29 compared five treatment subgroups of 240 patients. The group of trifusal, a potent inhibitor of cyclooxygenase, combined with low-intensity anticoagulation with vitamin K antagonists, showed fewer haemorrhages and embolic events than the group of conventional anticoagulation in isolation in patients with moderate risk. However, the small sample size of this study questions these results.
The Stroke Prevention Using Oral Thrombin Inhibitor in Atrial Fibrillation (SPORTIF) trial showed that ximelagatran, a direct thrombin inhibitor,30,31 reduced the incidence of embolic (1.6 versus 2.3% per year; p=not significant) and bleeding (25.8 versus 29.8% per year; p=0.007) events compared with warfarin. Unfortunately, this drug was associated with a transient and unpredictable increase of liver enzymes and was not finally approved for clinical use.
Other trials, such as The Atrial Fibrillation Trial of Monitored, Adjusted Dose Vitamin K Antagonist, Comparing Efficacy and Safety with Unadjusted SanOrg 34006/Idraparinux (AMADEUS), have evaluated idraparinux, a direct factor Xa inhibitor with a prolonged half-life that allows a single weekly subcutaneous injection for full anticoagulation. However, the high incidence of haemorrhage (19.7 versus 11.3%) was responsible for the early termination of this study.32 The Evaluation of Weekly Subcutaneous Biotinylated Idraparinux Versus Oral Adjusted-Dose Warfarin to Prevent Stroke and Systemic Thromboembolic Events in Patients with Atrial Fibrillation (BOREALISAF) trial was designed to re-evaluate this drug in 9,600 patients. The drug doses were adjusted according to patient age and the drug molecule had a bioetil added that allowed the use of an antidote, the avidine, in case of serious bleeding.33 Nevertheless, the study was discontinued by the sponsor following publication of the good results with other new anticoagulants.
Antiplatelets and Anticoagulation Combination
Combination of antiplatelets with anticoagulants is aimed at reducing or at least not increasing thromboembolic events but with a lower level of anticoagulation that may reduce haemorrhagic complications. Several studies have shown that addition of aspirin to vitamin K antagonists provides no added benefit to the latter in isolation.34,35 The already-mentioned NASPEAF trial29 is among the few to suggest a benefit of this combination in both high- and medium-risk patients. This makes combination of antiplatelets with anticoagulants a treatment that should not be used except for those patients with high embolic risk (i.e. recurrent embolism despite proper anticoagulation) and low bleeding risk.
The guidelines of the American College of Cardiology (ACC), the American Heart Association (AHA) and the European Society of Cardiology (ESC) provide a simplified approach for thromboembolic prophylaxis in patients with non-valvular AF according to their risk profile.7 Aspirin should be the drug of choice for patients with no or weak risk factors. Warfarin is recommended for patients with a single moderate risk factor, although aspirin should also be considered an alternative in these patients. Full anticoagulation is considered mandatory for patients with more than one moderate risk factor or one high risk factor (see Figure 2).
Despite these general recommendations, other factors, such as the likelihood of following correct anticoagulation control and the patient’s haemorrhagic risk profile, always have to be considered. In addition, different clinical situations merit special consideration. Patients with mechanical prosthetic heart valves require a higher level of anticoagulation, with an international normalised range (INR) between 2.5 and 3.5.6 Cardioversion also implies a special risk of embolism specially if AF lasts for more than 48 hours due to electrical remodelling and delay of recovery of atrial mechanical function. This is the reason for the current recommendation of full anticoagulation for at least three weeks before cardioversion and a minimum of four additional weeks following.
Problems of Anticoagulation with Vitamin K Antagonists
Vitamin K antagonists have been the only option for oral anticoagulation for the last 50 years. The benefits of these drugs have been discussed above. However, these drugs have manypharmacodynamic and pharmacokinetic problems.6,36 Their half-life is prolonged, the therapeutical window is very narrow and the dose–effect rate is highly variable because several factors influence the level of anticoagulation at the same dose. Among these latter factors the most important are: age, diet, stress, infection and other concomitant diseases, cytochrome P450 CYP2C9 and VKORC1 genetic polimorphysms, and concurrent medication. All of this means that frequently patients are not anticoagulated within the therapeutic range for long periods of time, which makes regular control mandatory for dose adjustment. Figure 3 presents the relationship between the anticoagulation level and the risk of thromboembolic and haemorrhagic complications.
An INR lower than two adds little protection and higher than three is associated with a stepped increase in cerebral haemorrhage without a significant additional decrease of thromboembolic events. This observation was shown in several independent studies.37 In addition, the proportion of time within the therapeutic range is inversely correlated with the number of strokes, either ischaemic or haemorrhagic.38 The mean proportion of time within the therapeutic range varies from country to country and even from institution to institution. The quality of control has improved dramatically in the last two decades as a result of the use of the INR, the generalisation of specialised clinics for INR control and the higher education of patients and the population in general. However, despite all these achievements and for the reasons explained above, the time outside the therapeutic range is over 50% in most patients under the best conditions.38,39 Fear about poor anticoagulation stability is possibly responsible for the high proportion of patients (around 20–50%) who do not receive this treatment despite a clear indication for it.40,41 Personal INR meters have been developed to facilitate and improve anticoagulation control. Some small trials suggested better control and a lower rate of thromboembolic and haemorrhagic complications. However, these results have not been validated by others and also require the competency and collaboration of the patient.42 More recently, genetic characterisation of patients on vitamin K antagonists, especially of cytochrome P450, CYP2C9 and VKORC1 genes, may improve the individual dose adjustment of warfarin. These genes are linked to the sensitivity of the anticoagulant effect of warfarin. However, the clinical impact of this strategy has not yet been evaluated.43,44
New Strategies for Thromboembolic Prevention
The problems of anticoagulation with vitamin K antagonists have led to the investigation of new drugs that can be administered orally and have better a dose–response relationship, shorter half-life and, especially, higher efficacy and safety without the need for frequent anticoagulation controls. The families of drugs more investigated in patients with AF are inhibitors of the activated coagulation factor X, including odipracil, rivaroxaban, apixaban and edoxaban, and inhibitors of coagulation factor II (thrombin), including ximelagatran and dabigatran. Table 2 summarises the main clinical trials with these new drugs in patients with AF.
Some of them, such those of ximelegatran and idraparinux, have been discussed above and the results of the rest have not yet been reported with the exception of those studying dabigatran. In addition, non-pharmacological therapies have been developed to prevent recurrent embolism in certain patient populations and some of them will be succinctly discussed at the end of this section.
The Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial is the first showing a drug with higher efficacy and safety than warfarin in patients with non-valvular AF.45 This non-inferiority trial enrolled 18,113 patients with non-valvular AF. Patients were randomly divided into three groups to receive two different doses of dabigatran (110 or 150mg twice a day) or warfarin, which was titrated according to non-blinded INR. The primary objective of the study was to prove non-significant differences in stroke or systemic embolism incidence during a mean follow-up of two years. The trial found that annual incidence of stroke or systemic embolism was 1.69% in the warfarin group, 1.53% in the 110mg dabigatran group and 1.11% in the 150mg dabigatran group, which represented a 9% and 34% reduction of RR of the lower (RR 0.91; CI 0.74–1.11; p< 0.001 for non-inferiority) and higher (RR 0.66; CI 0.53–0.82; p<0.001 for superiority) dabigatran doses, respectively. This effect was consistently observed in different patient subgroups, including those with prior stroke. The annual incidence of major haemorrhages was 3.36% in the warfarin group, similar and higher to those observed with the higher (3.11%, p=not significant) and lower (0.12%, p<0.001) dabigatran doses, respectively, and higher to that developed in the lower dabigatran dose group. In summary, 110mg dabigatran twice a day requires no anticoagulation controls and is associated with similar stroke and systemic embolism rates and a less-severe haemorrhage rate than warfarin. In addition, 150mg dabigatran twice a day is associated with lower stroke and systemic embolism rates but the severe haemorrhage rate is similar to warfarin. Cerebral haemorrhages are less common with both doses of dabigatran than with warfarin. The trial also showed a trend that reached statistical significance (2.28 versus 2.69%; p<0.04) in the 150mg group towards lower cardiovascular and total mortality in patients treated with dabigatran than in those with warfarin. These results make dabigatran an efficacious and safe alternative for patients with non-valvular AF and moderate–high risk of systemic thromboembolism. The extrapolation of these results to high-risk patients, such as those with a disease or prosthetic heart valve, warrants further investigation.
Currently, dabigatran is approved by most drug regulatory agencies for venous thrombosis prevention following orthopaedic surgery. However, thromboembolic prevention in AF patients is not yet accepted by most. Nevertheless, dabigatran should be considered in special situations such in patients with dicumarin allergy, poor INR control or recurrent haemorrhages or embolism under adequate vitamin K antagonist therapy.
Antithrombotic therapy is not the only strategy for thromboembolic prevention in patients with AF. Conversion and maintenance of sinus rhythm by catheter ablation should be followed by temporary anticoagulation but some have proposed permanent anticoagulation withdrawal in patients with no AF recurrence. Nevertheless, current evidence does not support this practice since no trial has demonstrated its safety and no differences in thromboembolic complications have been found between rhythm and heart rate control strategies in antiarrhythmic drug trials.
Left atrial appendage exclusion or resection at cardiac surgery is an effective method to prevent future embolic events and should be performed routinely in all those undergoing cardiac valve surgery.46 Similarly, percuteous left atrial appendage closure is aimed at abolishing the main source of cardiac embolic events. A device for percutaneous left atrial appendage closure showed a reduction in thromboembolic events compared with conventional treatment in a randomised trial Embolic Protection in Patients with Atrial Fibrillation ([PROTECT-AF]). However, the use of the device was associated with a significant number of periprocedural complications.47 Improvements of the device design, more experience with its use and better patient selection will surely establish the role of this technique in the management of high-risk patients with AF.
Intense research on thromboembolic profilaxis in patients with AF will surely change clinical practice in the near future. New anticoagulants will provide a more stable, efficacious and safe anticoagulation and lower therapy control burden for patients. Percutaneous techniques, such as ablation and left atrial appendage closure, will possibly decrease the need, time and intensity of antithrombotic drug treatment. Finally, better cardiovascular disease prevention will decrease AF incidence and thromboembolic risk factors.48