PDF: ATTICUS
(2023)Objective
Determine if apixaban is superior to aspirin in reducing new ischemic lesions in patients with enriched embolic stroke of undetermined source (ESUS) and high cardioembolic risk.
Study Summary
• In patients with recent ESUS and enrichment factors for cardioembolism (e.g., atrial cardiopathy), apixaban was not superior to aspirin with cardiac monitoring in preventing new ischemic brain lesions. The trial was terminated early for futility.
Intervention
Apixaban 5 mg BID (or 2.5 mg BID if dose-reduction criteria met) vs. aspirin 100 mg daily. Randomized 1:1 in open-label, blinded-endpoint design. All patients had at least one enrichment factor (e.g., high CHA₂DS₂-VASc score, atrial high-rate episodes, large LA, or PFO). MRI follow-up over 12 months.
Study Design
Arms: Apixaban 5 mg BID (n=178) vs Aspirin 100 mg daily (n=174)
Outcome
• Primary outcome (any new ischemic lesion on MRI during 12-month follow-up): 13.6% (apixaban) vs. 16.0% (aspirin); adjusted OR 0.79 (95% CI 0.42–1.48); P=0.57
• Recurrent stroke/systemic embolism: 5.7% vs. 6.4%; HR 0.81 (95% CI 0.36–1.80) — Fine-Gray
• Major or CRNM bleeding: 2.9% vs. 4.2%; HR 0.68 (95% CI 0.22–2.16) — Fine-Gray
• Atrial fibrillation detection: 22.9% (apixaban) vs. 27.6% (aspirin); HR 0.78 (95% CI 0.51–1.19)
• No intracranial hemorrhage occurred in either group
• Recurrent stroke/systemic embolism: 5.7% vs. 6.4%; HR 0.81 (95% CI 0.36–1.80) — Fine-Gray
• Major or CRNM bleeding: 2.9% vs. 4.2%; HR 0.68 (95% CI 0.22–2.16) — Fine-Gray
• Atrial fibrillation detection: 22.9% (apixaban) vs. 27.6% (aspirin); HR 0.78 (95% CI 0.51–1.19)
• No intracranial hemorrhage occurred in either group
Bottom Line
Apixaban was not superior to cardiac monitoring–guided aspirin in reducing new ischemic lesions in an enriched ESUS population.
Major Points
- ATTICUS tested apixaban vs aspirin in an ENRICHED ESUS population — patients with specific risk factors for cardioembolism (CHA2DS2-VASc ≥4, atrial high-rate episodes, LA abnormalities, or PFO). Published NEJM Evidence 2023.
- Primary outcome (any new ischemic lesion on MRI during 12-month follow-up): 13.6% vs 16.0% (adjusted OR 0.79, 95% CI 0.42-1.48, p=0.57) — no significant difference. Trial stopped early for futility.
- Distinguished from NAVIGATE ESUS and RE-SPECT ESUS by using an ENRICHMENT strategy — only included ESUS patients most likely to have occult cardioembolism, rather than all-comers.
- Remarkably high AF detection rate: 22.5-28.2% during 12 months of cardiac monitoring — much higher than the ~5-10% in typical ESUS populations, validating the enrichment approach for identifying occult AF.
- The aspirin arm had a BUILT-IN crossover: patients switched to apixaban upon AF detection. This design feature means the control arm became progressively more similar to the treatment arm over time.
- Zero intracranial hemorrhages in either group despite early anticoagulation initiation (3-28 days post-stroke) — strong safety signal for early DOAC use post-stroke.
- Only 352 patients enrolled (planned maximum 600 under an adaptive design) before futility stopping — severely underpowered for clinical endpoints. The MRI surrogate endpoint was novel but not validated.
- Challenges the ESUS concept: if a large portion of enriched ESUS patients develop AF during monitoring, then these are NOT truly cryptogenic — they are undiagnosed AF-related strokes.
- Supports the current paradigm shift toward PROLONGED cardiac monitoring rather than empiric anticoagulation for ESUS patients.
- Informed by failures of NAVIGATE ESUS (rivaroxaban, more bleeding, no benefit) and RE-SPECT ESUS (dabigatran, no benefit) — all three trials suggest empiric anticoagulation for unselected ESUS is not the answer.
Study Design
- Study Type
- Multicenter, randomized, open-label, blinded-outcome trial
- Randomization
- Yes
- Blinding
- Blinded outcome assessment
- Sample Size
- 352
- Follow-up
- 12 months
- Centers
- 16
- Countries
- Germany
Primary Outcome
Definition: Any new ischemic lesion on brain MRI during 12-month follow-up
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| 16.0% (25/156) | 13.6% (23/169) | 0.79 (0.42 to 1.48) | 0.57 |
Limitations & Criticisms
- Open-label design (PROBE) — physicians and patients knew the treatment assignment, potentially influencing concomitant medication use and monitoring intensity.
- Severely underpowered — only 352 of a planned maximum of 600 patients (adaptive design) enrolled before futility stopping. The small sample means even moderate treatment effects could not be detected.
- Surrogate primary endpoint (new MRI lesions at 12 months) is NOT a validated surrogate for clinical stroke recurrence — silent ischemic lesions have uncertain clinical significance.
- Built-in crossover design undermines the comparison: aspirin-arm patients who developed AF were switched to apixaban, making the control arm progressively more similar to the treatment arm.
- Short follow-up (12 months) may miss delayed recurrent events — many ESUS patients have low annual recurrence rates (2-5%), requiring longer observation.
- Single-country trial (Germany) with only 16 centers — limited generalizability to other populations and healthcare systems.
- The enrichment criteria, while innovative, selected a heterogeneous population — combining PFO patients with those having atrial high-rate episodes is mixing very different pathophysiologies.
- Industry-funded (Bristol-Myers Squibb, the apixaban manufacturer + Medtronic for cardiac monitors) — potential conflicts in trial design and futility stopping rules.
- No comparison with longer-duration cardiac monitoring alone (which might have detected AF in many patients, enabling targeted anticoagulation without empiric treatment).
Citation
Geisler T, Keller T, Martus P, et al. Apixaban versus Aspirin for Embolic Stroke of Undetermined Source. NEJM Evid. 2023;3(1).