Dan Atar, MD, PhD, FESC, FACC, FAHA
Non-vitamin K antagonist oral anticoagulant (NOAC) treatment can reduce the risk of stroke in patients with atrial fibrillation (AF). However, patients who take NOACs are at an increased risk of bleeding. For this reason, risk scores such as the HAS-BLED score have been developed to estimate the likelihood of major bleeding in patients with AF. For example, a HAS-BLED score of ≥3 indicates a high risk of major bleeding (Table 1).
Table 1. HAS-BLED score for the estimation of the risk of major bleeding.
The phase III trials on the prevention of stroke in patients with AF with NOACs (apixaban, dabigatran, edoxaban and rivaroxaban) enrolled different proportions of patients with a HAS-BLED score ≥3. As shown in the Figure, over half of the patients in ROCKET AF (rivaroxaban) and almost half of the patients in ENGAGE AF-TIMI 48 (edoxaban) had a HAS-BLED score ≥3.
Figure. HAS-BLED score patient distribution across major phase III trials of NOACs shown as the percentage of each study population with a HAS-BLED score ≥3. AF, atrial fibrillation; NOAC, non-vitamin K antagonist oral anticoagulant.
A prospective phase IV study (XANTUS) evaluated the effectiveness and safety of rivaroxaban in patients with AF in a real-world setting. Patients in XANTUS had a lower mean HAS-BLED score than those in ROCKET AF (2.0 vs 2.8, respectively), and this was reflected in the lower rate of major bleeding in XANTUS versus ROCKET AF (2.1 vs 3.6 events per 100 patient-years, respectively). However, it is important to note that adjustment for differences in baseline characteristics is necessary before comparing event rates between these studies, as shown in a recent analysis. Even though the mean HAS-BLED and CHADS2 scores in XANTUS were lower than those in ROCKET AF, a large proportion of patients in XANTUS were at high risk of bleeding (27.4% had a HAS-BLED score ≥3) and stroke (59.3% had a CHADS2 score ≥2).
When evaluating whether to prescribe a NOAC to a patient, it is important to consider their risk of stroke and any contraindications as per the label. In addition, their risk of bleeding should be assessed and reduced if possible, but a high bleeding risk score should not necessarily preclude the use of a NOAC to reduce the risk of stroke. The risk factors included in the HAS-BLED score overlap with several elements of the CHA2DS2-VASc score (Table 2), meaning that patients with a high risk of bleeding may also have a high stroke risk. Therefore, the high risk of bleeding in some patients will often be outweighed by a more pressing need to address a high risk of stroke. For this reason, the 2016 European Society for Cardiology (ESC) guidelines for the management of AF do not recommend the use of bleeding risk scores to identify patients who are unsuitable for NOAC treatment, but rather suggest that they should be used to identify risk factors that could be modified.
Table 2. HAS-BLED and CHA2DS2-VASc scores for the estimation of risk of major bleeding.
Consider, for example, a 75-year-old woman who is taking a non-steroidal anti-inflammatory drug and has a history of uncontrolled hypertension, poorly managed renal impairment and diabetes, and a high risk of bleeding and stroke (HAS-BLED score of 4 and a CHA2DS2-VASc score of 5). Several of these risk factors are modifiable. If her blood pressure is reduced to target levels and her renal impaCirment is optimally controlled, she will have a moderate risk of bleeding (HAS-BLED score of 2) and a high risk of stroke (CHA2DS2-VASc score of 4), and would clearly benefit from anticoagulant therapy.
When patients experience bleeding events during treatment with NOACs, it is important to consider the fact that new-onset organ-specific bleeding can be the first sign of an underlying malignancy in the same organ system. In these cases, further evaluation may be warranted to allow the prompt detection and treatment of any malignancy that may be present. This is consistent with the findings of a recent post hoc analysis of the COMPASS trial, which demonstrated an association between gastrointestinal or genitourinary bleeding and a subsequent diagnosis of gastrointestinal or genitourinary cancer, respectively, in patients receiving long-term antithrombotic treatment.
In summary, bleeding risk scores should not be looked at in isolation, and the overall net clinical benefit of anticoagulants should be considered when making treatment decisions in patients with AF. Any modifiable bleeding risk factors (e.g. uncontrolled hypertension, poorly managed hepatic or renal impairment, excessive alcohol consumption, and other medications) should be identified and corrected so that the overall risk of bleeding can be reduced.