Bleeding risk during oral anticoagulation therapy for atrial fibrillation

This editorial refers to ‘Major bleeding in patients with atrial fibrillation receiving vitamin K antagonists: a systematic review of randomized and observational studies’ by N.S. Roskell et al., on page 787. Oral anticoagulation (OAC) is highly effective at reducing the risk of stroke and mortality, but bleeding is the main complication of OAC treatment, which therefore diminishes its net clinical benefit. The fear of bleeding among physicians leads to lack of prescription of oral anticoagulants, and reduces uptake, in patients who may otherwise benefit from them.1,2 Stroke risk is closely related to bleeding risk in atrial fibrillation (AF) patients.3,4 Many risk factors for thromboembolism such as advanced age, uncontrolled hypertension, history of ischaemic heart disease, cerebrovascular disease, or previous bleeding events, have also been identified as risk factors for bleeding.5,6 Further, bleeding risk is inherent to anticoagulants mechanism of action, but particularly among vitamin K antagonists (VKAs), with numerous food, drug, and alcohol interactions, inter- and intra-patient dose–response variability, and narrow therapeutic window, which complicate management, and require close monitoring to reduce the risk of adverse events.6 Reducing major bleeding represents an important step in improving the net clinical benefit of OAC, therefore bleeding events have been systematically identified as a major endpoint in the assessment of safety during clinical trials, particularly those developing new OAC drugs. While for thrombotic assessment several risk stratification scores have been developed, and there is a general consensus on endpoint definitions, bleeding risk remains less clearly defined, often resulting in over-estimation.2 When approaching bleeding risk, a substantial heterogeneity remains among the many bleeding definitions currently used as well as lack of uniformity in bleeding criteria.6 This lack of standardization makes it difficult to optimally compare and interpret different results across studies, because findings may vary according to the definition of bleeding employed. The majority of studies use the International Society on Thrombosis and Haemostasis criteria for major bleeding,7 which combines both laboratory and clinical parameters, but even this classification suffers from several limitations, such as the subjectivity of a diagnosis of non-major but clinically relevant bleeding, or the decisions regarding the need for blood transfusion. Also, many of the events counted as major bleeds in clinical trials are clinically less significant and it may be helpful to distinguish them into either clinically relevant or less relevant major bleeds.6 The bleeding risk associated with VKA is well-known, and many previous studies have tried to assess major bleeding and mortality risk associated with such therapy. In the present issue of the Journal, Roskell et al.8 report the results of a systematic review of major bleeding in AF patients receiving VKA, using data from both randomized clinical trials and ‘real-life’ observational studies. The authors reviewed over 3000 records, 47 of them included in the systematic analysis (16 clinical trials and 31 observational studies). They found a higher incidence of major bleeding among observational studies (weighted mean of 4.4% per patient-year, with a greater variation in reported results) when compared with clinical trials (2.8% per patient-year).8 Of note, some data were either missing or not well-recorded in the original articles, with a lack of information about co-morbidities or concomitant medications, such as anti-platelet drugs, which may also influence the rate of both bleeding and mortality. The interpretation of main findings from this systematic review8 is also hampered by the significant clinical and methodological heterogeneity of the data: differences in the definitions of outcomes across the different studies, patient characteristics [inception cohorts (warfarin experienced vs. warfarin-naïve); higher risk of adverse events when initiating OAC; stroke/bleeding risk profiles at baseline]; and quality of follow-up data (e.g. closer monitoring and control of anticoagulation in clinical trials and greater physician awareness of major bleeding and increased event reporting). This considerable heterogeneity does not allow the authors to calculate rates based on a substantially higher absolute number of events which may have provided evidence for more robust conclusions, as a variety of events should have been pooled under the same term of ‘bleeding’. Interestingly, higher rates of bleeding events have been found over the time, both in the clinical trials and in observational studies, although with a bigger increase seen among observational studies.8 This may be partly explained by the widespread use of OAC over the past decade, given the proved net clinical benefit of OAC, as well as increasing indications leading to more likely treatment of elderly patients (with greater co-morbidities and poly-pharmacy) and better awareness of physicians, who are more likely to report OAC treatment-related bleeding events. One of the major limitations of this systematic review is the exclusion of smaller studies, those with <300 patients receiving VKA. This may have lead to the exclusion of many observational ‘real-life’ cohorts, evident by the fact that no observational study conducted before 2001 was included. Further, using this sample size cut-off they have also excluded many of the earlier clinical trials where VKA were compared with placebo/aspirin/low-dose VKA which show a greater net benefit for VKA therapy.9–11 We should not forget that the inherent limitations of the VKA therapy have promoted the development of novel oral anticoagulants (NOACs), which may offer efficacious and safer alternatives (with appropriate patient selection).12 However, their safety regarding long-term outcomes continues to be defined and extrapolating findings from randomized controlled trials to daily clinical practice remains difficult. Safety information from ‘real-life’ patients treated with both VKAs and NOACs is needed. Although the NOACs offer greater choice and availability of OAC to more AF patients, they may not be suitable for everyone and for some VKAs will still remain the only viable alternative. In the meantime, decisions regarding OAC prescription, whether with a VKA or a NOAC should be based on a patient's individual risk of stroke (as assessed by CHA2DS2-VASc) and tailored by their risk of bleeding (assessed by HAS-BLED).13,14 With the different NOACs available, clinicians now have a choice—and can match the drug profile to the particular patient's characteristics and associated co-morbidities. Greater focus needs to be placed on patient education so that they understand which factors under their control might affect the efficacy and safety of OAC, particularly medication adherence (as well as diet, alcohol, other medications, including over-the-counter drugs), and what signs and symptoms of bleeding they need to be aware of. Conflict of interest: G.Y.H.L. has served as a consultant for Bayer, Astellas, Merck, AstraZeneca, Sanofi, BMS/Pfizer, Daiichi-Sankyo, Biotronik, Portola and Boehringer Ingelheim and has been on the speakers bureau for Bayer, BMS/Pfizer, Boehringer Ingelheim, and Sanofi Aventis. D.A.L. is in receipt of an investigator-initiated educational grant from Bayer Healthcare and Boehringer Ingelheim and has been on the speakersã bureau for Boehringer Ingelheim, BMS/Pfizer, and Bayer Healthcare. She is also a Steering Committee member for the AEGEAN study. P.G. has no conflicts of interest to declare.