ESPRIT
(2006)Objective
Determine whether adding dipyridamole 200 mg twice daily to aspirin reduces vascular events versus aspirin alone in patients with TIA or minor ischaemic stroke of presumed arterial origin.
Study Summary
• Absolute risk reduction 1.0% per year (95% CI 0.1-1.8); NNT 104 per year (95% CI 55-1006)
• All major ischaemic events: HR 0.81 (95% CI 0.65-1.01); vascular death, stroke, or MI: HR 0.78 (95% CI 0.63-0.97)
• Major bleeding numerically lower with combination (35 vs 53; HR 0.67, 95% CI 0.44-1.03); discontinuation 34% vs 13% (mainly headache)
• Meta-analysis with 5 prior trials: overall RR 0.82 (95% CI 0.74-0.91) for vascular death/stroke/MI
Intervention
Aspirin 30-325 mg daily plus dipyridamole 200 mg twice daily (extended-release preferred)
Inclusion Criteria
Adults referred within 6 months of TIA (including transient monocular blindness) or minor ischaemic stroke (modified Rankin ≤3) of presumed arterial origin
Study Design
Arms: Aspirin + dipyridamole vs aspirin alone (open-label; blinded outcome adjudication)
Patients per Arm: 1363 aspirin+dipyridamole vs 1376 aspirin (total 2739)
Outcome
• Vascular death: 3.2% vs 4.4% (HR 0.75, 95% CI 0.51-1.10); all-cause death 6.8% vs 7.8% (HR 0.88, 95% CI 0.67-1.17)
• First ischaemic stroke: 7.0% vs 8.4% (HR 0.84, 95% CI 0.64-1.10); first cardiac event: 3.2% vs 4.4% (HR 0.73, 95% CI 0.49-1.08)
• Major bleeding: 2.6% vs 3.9% (HR 0.67, 95% CI 0.44-1.03); minor bleeding 12.5% vs 12.2% (RR 1.03)
• 34% of combination-arm patients discontinued vs 13% on aspirin — headache was the leading reason
Clinical Question
In patients with a TIA or minor ischaemic stroke of presumed arterial origin, does adding dipyridamole 200 mg twice daily to aspirin reduce major vascular events (composite of vascular death, non-fatal stroke, non-fatal MI, or major bleeding) compared with aspirin alone?
Bottom Line
Adding extended-release dipyridamole 200 mg twice daily to aspirin reduced the composite of vascular death, non-fatal stroke, non-fatal MI, or major bleeding by 20% (HR 0.80, 95% CI 0.66-0.98; ARR 1.0%/yr, NNT 104/yr) in patients with recent TIA or minor stroke of arterial origin; combined with ESPS-2 in an updated meta-analysis (RR 0.82, 95% CI 0.74-0.91) this established aspirin+dipyridamole as preferred antiplatelet therapy over aspirin monotherapy for secondary prevention after non-cardioembolic cerebral ischaemia — at the cost of a much higher discontinuation rate driven by dipyridamole-induced headache (34% vs 13%).
Major Points
- Primary composite (vascular death, non-fatal stroke, non-fatal MI, or major bleeding) occurred in 173/1363 (13%) on aspirin+dipyridamole vs 216/1376 (16%) on aspirin alone; HR 0.80 (95% CI 0.66-0.98).
- Absolute risk reduction 1.0% per year (95% CI 0.1-1.8), corresponding to NNT of 104 per year (95% CI 55-1006) to prevent one primary event.
- Effect was consistent across prespecified subgroups (age, sex, randomisation scheme, ischaemic heart disease history, large vs small vessel disease, Asian vs non-Asian; smallest interaction p=0.18) and across post-hoc subgroups of aspirin dose (<40, 40-100, >100 mg) and time from event.
- Updated meta-analysis pooling ESPRIT with 5 previous trials (ESPS-2, ACCSG-II, AICLA, Guiraud-Chaumeil, Kaye) yielded an overall RR of 0.82 (95% CI 0.74-0.91) for vascular death/stroke/MI across 7795 patients.
- Major bleeding was numerically LOWER with combination therapy (35 vs 53 events; HR 0.67, 95% CI 0.44-1.03), a finding the authors attribute to chance; minor bleeding was similar (RR 1.03).
- Tolerability was the main drawback: 470/1363 (34%) of combination patients discontinued vs 184/1376 (13%) on aspirin alone; dipyridamole-related headache was cited by 26% of those who stopped combination therapy.
- Design features that broaden generalisability: open-label, pragmatic dosing (aspirin 30-325 mg allowed), 79 hospitals in 14 countries, minimal exclusion criteria, and blinded outcome adjudication by independent committee.
- On-treatment analysis was similar to intention-to-treat (HR 0.82 vs 0.80), and observed event rate on aspirin (4.8%/yr) was lower than projected (6%/yr) but the trial's power was preserved by longer-than-planned 9722 patient-years of follow-up.
Study Design
- Study Type
- Randomized Controlled Trial (pragmatic, open-label with blinded outcome adjudication)
- Randomization
- Yes
- Blinding
- Open-label treatment; blinded (PROBE-style) outcome adjudication by independent auditing committee
- Sample Size
- 2739
- Follow-up
- Mean 3.5 years (SD 2.0); total 9722 patient-years; close-out July 1 - December 31, 2005
- Centers
- 79
- Countries
- Netherlands, UK, Singapore, Italy, Belgium, Spain, Sweden, France, Germany, Austria, Portugal, Australia, Switzerland, USA
Primary Outcome
Definition: Composite of death from all vascular causes, non-fatal stroke, non-fatal myocardial infarction, or non-fatal major bleeding complication — whichever event occurred first (intention-to-treat)
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| 216/1376 (16%) | 173/1363 (13%) | 0.8 (0.66-0.98) | <0.05 (CI excludes 1) |
Limitations & Criticisms
- Open-label design (treatment not blinded) could bias notification and reporting of soft outcomes, although outcome adjudication was blinded and only clinically significant events were counted.
- Wide latitude in aspirin dose (30-325 mg) with no fixed regimen; while intended to reflect real-world practice, it complicates direct comparison with fixed-dose trials such as ESPS-2 (aspirin 50 mg only).
- 83% of combination-arm patients used extended-release dipyridamole; the other 17% used immediate-release (with different bioavailability), potentially diluting effect estimates.
- Only ~30% of patients had large-vessel disease and only 66% had a completed minor stroke; two-thirds were randomised 1-6 months after the event, so ESPRIT does not address the acute high-risk window in which subsequent trials (CHANCE, POINT, THALES) focus.
- Observed primary event rate on aspirin alone (4.8%/yr) was lower than the 6%/yr assumed for power, reflecting concomitant statin and antihypertensive use — leading to reduced absolute benefit and a wide NNT confidence interval (55-1006).
- High discontinuation rate (34% vs 13%) driven by dipyridamole-induced headache limits real-world durability of benefit and biases the ITT estimate towards null; a titration strategy was not tested.
- Twenty-four patients from one hospital were excluded before analysis for incomplete data, and follow-up was truncated at 11 patients from four other hospitals — small but non-random exclusions.
- Classification of large vs small vessel disease relied on CT/clinical features (no routine diffusion-weighted MRI), so subgroup analyses by lesion type may have been misclassified in 10-20%.
- The trial was not designed to detect a bleeding difference, so the unexpected numerical reduction in major bleeding with combination therapy is best interpreted as chance rather than a true safety signal.
Citation
Lancet 2006;367(9523):1665-1673. DOI: 10.1016/S0140-6736(06)68734-5