Performance of the ABC Scores for Assessing the Risk of Stroke or Systemic Embolism and Bleeding in Patients With Atrial Fibrillation in ENGAGE AF-TIMI 48

Abstract

Background: The ABC (age, biomarker, clinical history)-stroke and ABC-bleeding risk scores incorporate clinical variables and cardiovascular biomarkers to estimate risk of stroke or systemic embolic events and bleeding, respectively, in patients with atrial fibrillation. These scores have been proposed for routine clinical use, but their performance in external cohorts remains uncertain.

Methods: ENGAGE AF-TIMI 48 (Effective Anticoagulation With Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction 48) was a multinational randomized trial of the oral factor Xa inhibitor edoxaban in patients with atrial fibrillation and a CHADS2 score ≥2. We performed a nested prospective biomarker study in 8705 patients, analyzing baseline high-sensitivity troponin T (hsTnT), NT-proBNP (N-terminal B-type natriuretic peptide), and growth differentiation factor-15 (GDF-15), as well as in serial samples after 12 months. The ABC-stroke (age, prior stroke/transient ischemic attack, hsTnT, NT-proBNP) and ABC-bleeding (age, prior bleeding, hemoglobin, hsTnT, and GDF-15) scores were tested. Hazard ratios were adjusted for estimated glomerular filtration rate and the components of the CHA2DS2-VASc and HAS-BLED scores, respectively. Discrimination and reclassification were compared with these established scores.

Results: Median baseline hsTnT, NT-proBNP, and GDF-15 levels were 13.7 ng/L (25th-75th percentiles, 9.6-20.4 ng/L), 811 pg/mL (386-1436 pg/mL), and 1661 pg/mL (1179-2427 pg/mL), respectively. Elevated hsTnT, NT-proBNP, and GDF-15 were independently associated with higher rates of stroke or systemic embolic events, and elevated hsTnT and GDF-15 were independently associated with higher rates of major bleeding ( P<0.001 for each). The ABC-stroke and ABC-bleeding scores were well calibrated and yielded higher c indexes than the CHA2DS2-VASc score for stroke or systemic embolic events (0.67 [95% CI, 0.65-0.70] versus 0.59 [95% CI, 0.57-0.62]; P<0.001) and HAS-BLED score for major bleeding (0.69 [95% CI, 0.66-0.71] versus 0.62 [95% CI, 0.60-0.64]; P<0.001), respectively. The ABC-stroke and ABC-bleeding scores stratified patients within CHA2DS2-VASc and HAS-BLED risk categories ( P<0.001 for both). Patients with ABC-bleeding scores predicting a high 1-year risk of bleeding (>2%) derived greater benefit from treatment with edoxaban compared with warfarin.

Conclusions: The ABC-stroke and ABC-bleeding scores evaluated in this anticoagulated clinical trial cohort were well calibrated and outperformed the CHA2DS2-VASc and HAS-BLED scores, respectively. These scores may help identify patients most likely to derive a benefit from treatment with non-vitamin K antagonist oral anticoagulants.

Clinical trial registration: URL: https://www.clinicaltrials.gov . Unique identifier: NCT00781391.

Keywords: atrial fibrillation; biomarkers; hemorrhage; risk assessment; stroke.

Figures

Figure 1.. Cardiovascular biomarkers and annualized rate of stroke or systemic embolism and major bleeding.

High-sensitivity troponin T (hsTnT), N-terminal B-type natriuretic peptide (NT-proBNP), and growth differentiation factor-15 (GDF-15) were each independently associated with higher rates of S/SEE. hsTnT and GDF-15 were both independently associated with higher rates of bleeding, but NT-proBNP was not. GDF-15 indicates growth differentiation factor-15; hsTnT, high-sensitivity troponin T; HR, hazard ratio; Int, interval; ng/L, nanograms/liter; NT-proBNP, N-terminal B-type natriuretic peptide; pg/mL, picograms/milliliter.

Figure 2.. Relative importance of each variable in the risk models for stroke or systemic embolism and major bleeding.

The relative contributions were assessed using the partial Waldχ2 minus the predictor degrees of freedom. The components of the (A) ABC-stroke and (B) ABC-bleeding risk scores are bolded. AF indicates atrial fibrillation; CAD, coronary artery disease; CHF, congestive heart failure; df, degrees of freedom; GDF-15, growth differentiation factor-15; hsTnT, high-sensitivity troponin T; MI, myocardial infarction; NT-proBNP, N-terminal B-type natriuretic peptide; PAD, peripheral artery disease; TIA, transient ischemic attack.

Figure 2.. Relative importance of each variable in the risk models for stroke or systemic embolism and major bleeding.

The relative contributions were assessed using the partial Waldχ2 minus the predictor degrees of freedom. The components of the (A) ABC-stroke and (B) ABC-bleeding risk scores are bolded. AF indicates atrial fibrillation; CAD, coronary artery disease; CHF, congestive heart failure; df, degrees of freedom; GDF-15, growth differentiation factor-15; hsTnT, high-sensitivity troponin T; MI, myocardial infarction; NT-proBNP, N-terminal B-type natriuretic peptide; PAD, peripheral artery disease; TIA, transient ischemic attack.

Figure 3.. Annualized rates of (A) stroke or systemic embolic event stratified by the CHA2DS2-VASc and ABC-stroke risk scores and (B) major bleeding stratified by the HAS-BLED and ABC-bleeding risk scores.

Observed annualized event rates are shown as percent of patients at-risk. (A) The ABC-stroke risk score accurately stratified patients irrespective of CHA2DS2-VASc risk score categories (ptrend <0.001). (B) The ABC-bleeding risk score accurately stratified patients irrespective of HAS-BLED risk score categories (ptrend <0.001).

Figure 3.. Annualized rates of (A) stroke or systemic embolic event stratified by the CHA2DS2-VASc and ABC-stroke risk scores and (B) major bleeding stratified by the HAS-BLED and ABC-bleeding risk scores.

Observed annualized event rates are shown as percent of patients at-risk. (A) The ABC-stroke risk score accurately stratified patients irrespective of CHA2DS2-VASc risk score categories (ptrend <0.001). (B) The ABC-bleeding risk score accurately stratified patients irrespective of HAS-BLED risk score categories (ptrend <0.001).

Figure 4.. Cumulative incidence of stroke or systemic embolism and major bleeding stratified by 1-year risk estimates from the ABC scores.

(A) The observed cumulative incidence of S/SEE over the first three years following trial enrollment was stratified by the 1-year risk of S/SEE predicted by the ABC-stroke score (2%). The observed annualized event rates (shown for each risk group) were significantly different across strata (p2%). The observed annualized event rates (shown for each risk group) were significantly different across strata (p

Figure 4.. Cumulative incidence of stroke or systemic embolism and major bleeding stratified by 1-year risk estimates from the ABC scores.

(A) The observed cumulative incidence of S/SEE over the first three years following trial enrollment was stratified by the 1-year risk of S/SEE predicted by the ABC-stroke score (2%). The observed annualized event rates (shown for each risk group) were significantly different across strata (p2%). The observed annualized event rates (shown for each risk group) were significantly different across strata (p