Neurology Clinical Trial Database
COSS
Extracranial-Intracranial Bypass Surgery for Stroke Prevention in Hemodynamic Cerebral Ischemia
Clinical Question
Does extracranial-intracranial (EC-IC) bypass surgery, added to best medical therapy, reduce subsequent ipsilateral ischemic stroke in patients with recently symptomatic internal carotid artery occlusion (AICAO) and hemodynamic cerebral ischemia?
Bottom Line
EC-IC bypass surgery did not reduce the risk of ipsilateral ischemic stroke at 2 years compared with medical therapy alone.
Major Points
- EC-IC bypass did not reduce ipsilateral stroke rates compared to medical therapy (21.0% vs 22.7% at 2 years, P=0.78) — a definitively negative trial for surgical revascularization.
- Second major negative trial for EC-IC bypass after the original EC/IC Bypass Study (1985) — even with hemodynamic selection, surgery fails to prevent stroke.
- Used PET oxygen extraction fraction (OEF ratio >1.130) to select patients with true hemodynamic compromise — the most rigorous selection criteria ever applied, yet still failed to show benefit.
- Stopped early for futility after enrolling only 195 of planned 372 patients — the DSMB determined no reasonable chance of demonstrating benefit.
- 30-day perioperative stroke rate was 14.3% in the surgical group (all ipsilateral ischemic) — this unacceptably high complication rate negated any potential long-term benefit.
- Medical therapy arm had better-than-expected outcomes (22.7% vs projected 40% stroke rate) — modern medical management dramatically reduced stroke risk, making the surgical bar impossible to clear.
- High graft patency (98%) and improved OEF in surgical patients confirmed technical success — the surgery worked hemodynamically but did not prevent strokes, suggesting the mechanism of stroke in ICA occlusion is embolic rather than hemodynamic.
- Extremely slow enrollment: 195 patients over 8 years (2002–2010) across 49 centers — reflecting the rarity of patients meeting PET OEF criteria and limited clinical equipoise.
- Led to guideline-level abandonment of EC-IC bypass for atherosclerotic ICA occlusion — medical therapy is now the standard of care.
- Stimulated research into alternative approaches: encephaloduroarteriosynangiosis (EDAS) is being evaluated in the SIPS and JETS trials as potentially safer indirect revascularization.
Design
Study Type: Randomized, open-label, blinded-adjudication controlled trial
Randomization: 1
Blinding: Blinded adjudication only
Enrollment Period: 2002–2010
Follow-up Duration: 2 years
Centers: 49
Countries: USA, Canada
Sample Size: 195
Analysis: Intention-to-treat and on-treatment analysis with Kaplan-Meier estimates and z-tests
Inclusion Criteria
- Atherosclerotic internal carotid artery occlusion (AICAO)
- Hemispheric TIA or stroke within 120 days
- Hemodynamic ischemia confirmed by PET OEF ratio >1.130
- Suitable anatomy for EC-IC bypass
Exclusion Criteria
- Non-atherosclerotic carotid occlusion (dissection, vasculitis, radiation)
- TIA or stroke not in the territory of the occluded ICA or outside 120-day window
- Bilateral carotid occlusion
- Intracranial large vessel stenosis or occlusion requiring separate intervention
- Active malignancy or life expectancy <2 years
- Inability to provide informed consent or attend follow-up
- Medically unfit for surgery (severe cardiac, pulmonary, or systemic disease)
- Previous ipsilateral EC-IC bypass surgery
Arms
| Field | EC-IC Bypass + Medical Therapy | Control |
|---|---|---|
| Intervention | Microsurgical anastomosis of superficial temporal artery to MCA branch | Antithrombotics, risk factor control |
| Duration | 2-year follow-up | 2-year follow-up |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| All stroke and death within 30 days post-surgery and ipsilateral ischemic stroke within 2 years | Primary | 22.7% | 21.0% | 1.70% | 0.78 |
| Any stroke: 23.4% vs 26.9% | Secondary | 26.9% (95% CI 17.6-36.2); 24 events / 98 nonsurgical | 23.4% (95% CI 14.8-32.0); 22 events / 97 surgical | Difference 3.5% (95% CI -9.2 to 16.1) favoring surgery | P=0.59 |
| Fatal stroke: 1.0% vs 2.4% | Secondary | 2.4% (95% CI 0-5.6); 2 events / 98 nonsurgical | 1.0% (95% CI 0-3.1); 1 event / 97 surgical | Difference 1.3% (95% CI -2.5 to 5.2) favoring surgery | P=0.50 |
| Disabling stroke: 5.9% vs 2.4% | Secondary | 2.4% (95% CI 0-5.6); 2 events / 98 nonsurgical | 5.9% (95% CI 0.8-10.4); 5 events / 97 surgical | Difference -3.2% (95% CI -9.0 to 2.6) favoring nonsurgical | P=0.27 |
| Death: 1.0% vs 5.1% | Secondary | 5.1% (95% CI 0.2-9.9); 5 events / 98 nonsurgical | 1.0% (95% CI 0-3.1); 1 event / 97 surgical | Difference 4.0% (95% CI -1.2 to 9.7) favoring surgery | P=0.13 |
| Any stroke or death: 23.4% vs 29.9% | Secondary | 29.9% (95% CI 20.1-39.8); 26 events / 98 nonsurgical | 23.4% (95% CI 14.8-32.0); 22 events / 97 surgical | Difference 6.5% (95% CI -6.5 to 19.6) favoring surgery | P=0.33 |
| 30-day perioperative stroke in surgical group | Adverse | 14.3% | |||
| Perioperative stroke types | Adverse | All ipsilateral ischemic | |||
| Surgical complications | Adverse | Hematomas, MI, seizures, infections |
Criticisms
- 14.3% perioperative stroke rate was unacceptably high — negated any potential long-term surgical benefit and raised questions about surgical technique and patient selection.
- Severely underpowered due to early termination for futility (195 of 372 planned patients) — may have missed a smaller but real benefit in a more optimally selected subgroup.
- Historical control assumptions overestimated medical therapy stroke risk (projected 40%, observed 22.7%) — the trial was designed against an outdated baseline.
- Only PET OEF used for hemodynamic selection — MR perfusion, SPECT, or TCD vasoreactivity may identify different or overlapping populations that could benefit.
- Extremely slow enrollment (195 patients over 8 years across 49 centers) limits generalizability — only highly selected patients at specialized PET centers participated.
- Medical therapy improved dramatically during enrollment (2002–2010), including widespread statin use and tighter BP control, making the control arm progressively stronger.
- Open-label design — surgeons and patients knew allocation, potentially affecting post-operative management, stroke detection, and event reporting.
- Single bypass technique (STA-MCA anastomosis) — indirect revascularization approaches like EDAS may offer different risk-benefit profiles that COSS could not evaluate.
- No long-term follow-up beyond 2 years — potential late benefit from improved hemodynamics could not be assessed due to trial termination.
Funding
USPHS grants NS39526, NS42157, NS41895 (NINDS)
Based on: COSS (JAMA, 2011)
Authors: Powers WJ, Clarke WR, Grubb RL Jr, ..., Derdeyn CP
Citation: JAMA. 2011;306(18):1983–1992. doi:10.1001/jama.2011.1610
Content summarized and formatted by NeuroTrials.ai.