Bleeding and Risk for Future Cardiovascular Events in Patients With Atrial Fibrillation on Oral Anticoagulation

Major Bleeding Is a Major Problem

Josephine Harrington; Christopher B. Granger


Eur Heart J. 2022;43(47):4909-4911. 

Graphical Abstract

Bleeding and risk of subsequent events for patients with AF on anticoagulation. AF, atrial fibrillation; aHR, adjusted hazard ratio; LAA, left atrial appendage; MI, myocardial infarction; NSAID, non-steroidal anti-inflammatory drugs.

Oral anticoagulation therapies, and particularly direct-acting oral anticoagulants (DOACs), prevent over two-thirds of strokes for patients with atrial fibrillation (AF), and result in an estimated 34% reduction in mortality compared with no anticoagulation.[1] Despite this, anticoagulation is only used in about half of patients with AF and risk factors for stroke.[2] When anticoagulation is used, it is discontinued >20% of the time within 5 years.[3] The single most important reason for not using oral anticoagulants is concern over bleeding.[4] Thus, understanding the prevalence of bleeding, the predictors of bleeding, treatment after bleeding, and outcomes after bleeding in patients with AF is important to guide optimal use of anticoagulation.

Major bleeding occurs in ~4% of patients with AF on anticoagulation during the first year of follow-up in recent large trials.[5] Tools such as the HASBLED score can be used to stratify risk of bleeding, but fail to provide a high degree of discrimination. Furthermore, many of the risk factors associated with bleeding also represent risk factors for stroke, a fact that was highlighted when the CHADS scores were shown to be equally good at predicting both stroke and bleeding risk.[6] It can therefore be challenging to differentiate risk of stroke from risk of bleeding in a way that may improve risk–benefit assessment. The ABC bleeding risk score, which incorporates both clinical factors and predictive biomarkers, may be a more promising approach.[7] This risk model uses three readily available clinical factors—older age, prior bleeding, and lower haemoglobin—and two biomarkers, growth/differentiation factor 15 (GDF15) and high-sensitivity troponin, to stratify risk. Factors such as prior bleeding and lower haemoglobin and GDF15 are especially important since they predict risk of bleeding but not stroke, and may be useful to differentially identify bleeding risk.

In this issue of the European Heart Journal, Meyre and colleagues[8] used high quality Swiss registries of patients with AF to address an important question: what is the risk of adverse outcomes (either cardiovascular or recurrent bleeding) following bleeding events? In this analysis, they found that over 4 years, 20% of patients with AF on oral anticoagulation had a bleeding event, and 9% had a major bleeding event. The composite clinical outcome of stroke, myocardial infarction, or death from any cause occurred in nearly half of patients following major bleeding, at a median of 142 days after the bleeding event. Patients who had a major bleeding event were at twice the risk of a subsequent cardiovascular event, even after adjustment for other risk factors, in comparison with those without a bleeding event, but no significant difference in future cardiovascular events was seen between patients with clinically relevant non-major bleeding and patients without bleeding. Patients with a major bleeding event also had a shorter median time to a cardiovascular event than patients with clinically relevant non-major bleeding or without any bleeding. Importantly, discontinuation of anticoagulation occurred in 21% of patients following major bleeding, and those who discontinued were at more than twice the risk of a cardiovascular event compared with those who continued anticoagulation, though rates of cardiovascular events were also elevated for patients with major bleeding who continued their anticoagulation. Of patients with a new bleeding event, ~25% subsequently developed another bleeding event. However, only 4–5% of patients with bleeding had a subsequent major bleeding event, at similar rates regardless of whether the first event was a major or clinically relevant non-major bleeding event.

These findings from this study provide several important insights. First, bleeding is common among patients with AF on anticoagulation. Second, patients who experience bleeding events are at high risk of having their anticoagulation discontinued. Finally, patients who experience bleeding are at a high risk of subsequent stroke, myocardial infarction, or death, especially if they (i) experienced major bleeding or (ii) had discontinuation of their anticoagulation following their initial bleeding episode. Limitations to this analysis include: the modest number of patients in the analysis, which contributed to a small number of patients with an observed stroke outcome; the selectivity of the population due to need for provide consent to be in registries; and the fact that associations do not prove cause and effect.

What are the clinical implications of this study? While the focus of anticoagulation is to prevent a more important outcome than bleeding, namely stroke, this study provides novel information about the relationship between bleeding and subsequent risk of future death and cardiovascular events, including stroke and myocardial infarction (Graphical Abstract). Specifically, this study shows that patients who experience bleeding events are at a significantly higher risk for death and cardiovascular events.

Though it is not possible to define the reasons for this relationship, several factors are likely to contribute. First, it is possible that immediately after a bleeding event, activation of the coagulation cascade contributes to a prothrombotic state. The authors found that median time to cardiovascular event after bleeding was 142 days, even among patients who continued their anticoagulation, suggesting that the increased risk is not primarily due to the immediate impact of the bleeding episode. However, it is possible that there is at least some degree of increased risk immediately following a bleeding event, as was suggested in a subanalysis from the ARISTOTLE trial.[9] Second, risk factors for bleeding tend to also be risk factors for stroke and other cardiovascular events, so bleeding is a marker for having risk factors that increase their risk for stroke, such as renal dysfunction and advanced age. Finally, the authors noted that 21% of patients discontinued their anticoagulation following a major bleed, and that these patients had hazards of death, myocardial infarction, or stroke that were over three times higher than those of patients who continued their anticoagulation following their bleeding event. Therefore, patient and provider responses to bleeding (namely discontinuation of anticoagulation) probably also contribute to the higher risk of future stroke, myocardial infarction, and death observed after a bleeding episode. It is also likely that discontinuation of anticoagulation was a marker for greater concern of rebleeding and thus a marker of higher risk.

Taken together, these findings emphasize the need to minimize bleeding risk for patients with AF on anticoagulation. Though many of the shared risk factors for bleeding and stroke are not modifiable, the prevention of bleeding events may help to avoid precipitating a prothrombotic state. More importantly, avoiding bleeding events allows patients to stay on anticoagulation for stroke prophylaxis. Because patients with bleeding tend to also be those at the highest risk for stroke, this is particularly important.

There are many strategies to avoid bleeding, including avoidance of other drugs such as antiplatelet therapies and non-steroidal anti-inflammatory drugs, both of which are known to increase bleeding risk, use of therapies such as proton pump inhibitors to reduce bleeding when appropriate, blood pressure control, and use of DOACs, which have been associated with lower risk of bleeding compared with warfarin.[10] The use of risk scores to identify patients who are most likely to bleed, which include reversible factors, helps to guide decisions about which of these strategies should be employed. Studies have shown that strategies to avoid bleeding, such as avoiding antiplatelet therapy when using anticoagulants even for patients with (stable) coronary disease, may also reduce thrombotic events and mortality.[11]

Once a patient does experience a bleeding event, these findings highlight the importance of resuming anticoagulation when possible. The disproportionately elevated risk of death, myocardial infarction, and stroke following cessation of anticoagulation seen in this study, combined with relatively low rates of recurrent major bleeding when anticoagulation was resumed, even after an initial major bleed, suggest that for most patients with bleeding the benefits of resuming anticoagulation probably outweigh the risk of bleeding. Therefore, though anticoagulation should generally be stopped acutely during a major bleeding episode, re-initiation should be pursued whenever feasible.

An ongoing challenge for the healthcare provider is that patients at highest risk of bleeding are at highest risk of stroke. It is very clear that most high-risk patients, including the elderly, the frail at risk of falls, and those with renal impairment, have greater absolute benefit from anticoagulation, particularly with DOACs, even though the risk of bleeding is high. However, preventing bleeding is especially important in these populations. For intracranial haemorrhage, the most serious of all bleeding, using DOACs rather than warfarin is the most important opportunity to reduce risk.[5]

For patients with recurrent serious bleeding, and who refuse or are unable to resume anticoagulation, alternatives should be considered. The LAAOS 3 trial was the first large controlled trial to show superiority in stroke reduction with left atrial appendage (LAA) occlusion, albeit surgical occlusion on top of anticoagulation.[12] The improving safety of percutaneous LAA occlusion, with growing evidence of efficacy and safety, shows that LAA occlusion should be considered for stroke prevention for those not suitable for anticoagulation. Ongoing and future trials of new anticoagulants that may be safer, particularly those that inhibit factor XI, will determine whether these drugs may improve outcome for patients at high bleeding risk.[13–15]

In summary, this study by Meyre et al. provides important information on bleeding and its consequences in patients with AF on anticoagulation. These findings call for further investigation into strategies to reduce bleeding risk in these patients and to better understand how and when to resume anticoagulation after bleeding events.