In order to bring you the best possible user experience, this site uses Javascript. If you are seeing this message, it is likely that the Javascript option in your browser is disabled. For optimal viewing of this site, please ensure that Javascript is enabled for your browser.
Did you know that your browser is out of date? To get the best experience using our website we recommend that you upgrade to a newer version. Learn more.

We use cookies to optimise the design of this website and make continuous improvement. By continuing your visit, you consent to the use of cookies. Learn more

AMADEUS: Development of a novel composite stroke and bleeding risk score in patients with atrial fibrillation

Atrial Fibrillation

Presenter abstract
Discussant report
All the Scientific resources on ESC Congress 365


By Gregory Y H Lip
Other authors: Dr Dierdre A Lane, United Kingdom; Dr Harry Buller, Netherlands; Dr Stavros Apostolakis, United Kingdom
Given the overlap between stroke and bleeding risk factors, a risk stratification score for thromboembolism and serious bleeding could potentially be developed for patients with atrial fibrillation (AF). It is uncertain if such a composite score offers better predictive value or practical application against existing individual stroke and bleeding scores.

We used data from the vitamin K antagonist (VKA) arm of the AMADEUS trial, which was a multicentre, randomised, non-inferiority study that compared fixed-dose idraparinux with VKA in AF patients. We assessed predictors of outcome using two composite endpoints: Endpoint 1 was the sum of stroke or non-CNS systemic embolism and major bleeding. Endpoint 2 was defined as the sum of stroke, systemic or venous embolism, myocardial infarction, cardiovascular death or major bleeding.

The AMADEUS study randomized 2293 patients to the VKA arm. In a binary logistic regression analysis the independent predictors for composite endpoint 1 (50 events; 2.4/100 patient-years), were age (p=0.014), previous stroke/TIA (p=0.049), aspirin use (p=0.002) and time in therapeutic range (p=0.007).
For the composite endpoint 2 (94 events; 4.5/100 patient-years), similar predictors were evident, plus left ventricular systolic dysfunction (p=0.011).
We used these factors to develop scores by applying weight for each variable relative to the magnitude of the regression coefficients.
Composite score 1 = (0.05 x Age)+(0.6 x Previous stroke or TIA)+(0.9 x concomitant aspirin)-(1.8 x TTR)
Composite score 2 = (0.05 x Age)+(0.6 x Previous stroke or TIA)+(0.7 x concomitant aspirin)+(0.6 x LV dysfunction)-(1.4 x TTR)
The weighted scores (composite scores 1 and 2) demonstrated the highest numerical discriminatory performance (AUC=0.728; 95%CI: 0.659-0.798 and AUC=0.707; 95%CI: 0.655-0.758, for the endpoints 1 and 2 respectively), and positive net reclassification compared to all other scores.
The predictive or practical clinical utility of these new scores, when compared to the use of existing individual stroke and bleeding risk scores will be presented for the first time at the ESC meeting.
We have developed and validated a novel composite score for stroke/thromboembolism/bleeding.  Whilst individual risk stratification with existing stroke and bleeding scores would allow personalized balancing of the risk of stroke/thromboembolism against serious bleeding, the value of this composite score merits further study.

Discussant Report

Francisco Marin
Atrial fibrillation (AF) is associated with increased risk of stroke and thromboembolism, and thus oral anticoagulation (OAC) therapy is indicated in patients considered at moderate-to-high risk. The 2012 European Society of Cardiology (ESC) guidelines recommend the CHA2DS2-VASc score for stroke risk stratification in patients with AF, whereby following the initial identification of ‘truly low risk’ patients, those with ≥1 stroke risk factors can be offered effective stroke prevention, which is OAC.

The use of OAC is associated with an increased risk of major bleeding, with intracranial haemorrhage being the fearest complication associated with warfarin. The HAS-BLED bleeding risk score is the recommended score in the ESC guidelines as a practical tool to assess the individual bleeding risk in AF patients, whereby a score of ≥3 indicates ‘high risk’1. These risk stratification scales have been validated in different clinical scenarios, and have a potential role predicting different end-points, as all-cause death or other thrombotic complications, i.e. myocardial infarction.

The net clinical benefit favours anticoagulation for almost all AF patients with the exception of those at very low risk of ischemic stroke, essentially those with a CHA2DS2-VASc score of 0. Based on the results of a real-world nationwide cohort study of patients with non-valvular AF, the net clinical benefit was clearly in favour of VKA treatment in AF patients with increased risk of stroke/thromboembolism, regardless of bleeding risk estimation (as assessed by the HAS-BLED risk score).

In the present Clinical Trial Update, Lip et al. explore an interesting question, to create a composite risk score to assess the overall clinical outcome in AF, considering both stroke/thromboembolism and/or serious bleeding. They tested two composite clinical endpoints, that is:
(i) thromboembolism/major bleeding (a simple composite, Endpoint 1), and
(ii) stroke, systemic or venous embolism, myocardial infarction, cardiovascular death and major bleeds (a more complex composite, Endpoint 2). The study of Lip et al was based on a study population of 2,293 patients in the VKA arm of the AMADEUS trial.

The authors found that the independent predictors for Endpoint-1 were age, previous stroke/TIA, aspirin use and time in therapeutic range (TTR); and for Endpoint-2 were the same predictors plus left ventricular dysfunction. Based on this, two new composite risk scores were developed, which both offer good discriminative and predictive performance.  However, when the authors compared these new composite scores to the established risk scores, that is CHADS2, CHA2DS2-VASc or HAS-BLED, there were no significant improvements in the endpoint predictions.

The strengths of the study of Professor Lip et al. are remarkable

  • To the best of my knowledge it is the first formal assessment of a composite risk score to assess overall clinical outcome, taking into account the development of thromboembolic, bleeding and many often forgotten other cardiovascular events, such as myocardial infarction. 
  • The authors explored their hypothesis in a well defined cohort of AF patients, from the AMADEUS trial.
  • The authors confirmed various variables which have been proposed as important risk factors for adverse events in AF, such as concomitant aspirin intake and the quality of OAC (assessed by TTR). The addition of aspirin to VKA did not offer more protection but increased the risk of bleedings. Thus, we should check the quality of OAC control in our patients, and in the case of non-optimal INR control, look for another option (e.g. new oral anticoagulants)
  • The conclusion of Professor Lip in his presentation is very sensible:  ‘Simplicity is best’.  Indeed, we need simple tools for assessing the risk of our patients in the daily clinical practice for decision making: use a stroke score (CHA2DS2-VASc) to assess stroke risk, and a bleeding score (HAS-BLED) to assess bleeding risk.

The findings of the present study should be interpreted in the setting of the study design and its limitations, as the authors recognise. The authors included patients from a clinical trial, with their inherent limitations. All the patients were taking OAC, and we need to understand the global risk of our patients, with or without OAC. When considering the balance of risk and benefit, not every major complication (major bleeding, embolism, cardiac events) can be weighted equally, and patient’s values and preferences also should be considered.  

In conclusion, whilst Lip et al are to be commended for having validated 2 novel composite scores for stroke/thromboembolism/bleeding that offer good discriminatory and predictive performance, these composite risk scores did not perform better than the easier and more user-friendly ‘classical’ stroke and bleeding risk scores that are currently in use. Existing risk scores for stroke and bleeding separately offer far greater practically, and more personalised balancing of risks, as well as simplicity.   

Conflicts of interest
Dr. Marín has received research grants from and served in the speakers bureau of Bayer, Boehringer-Ingelheim and Pfizer/BMS.




AMADEUS: Development of a novel composite stroke and bleeding risk score in patients with atrial fibrillation

The content of this article reflects the personal opinion of the author/s and is not necessarily the official position of the European Society of Cardiology.