What were the key findings of the EVA-3S trial?

← Back

EVA-3S

Endarterectomy versus Stenting in Patients with Symptomatic Severe Carotid Stenosis

Share Share Copied! Compare

Year of Publication: 2006

Authors: Mas JL, Chatellier G, Beyssen B, et al. (EVA-3S Investigators)

Journal: N Engl J Med

Citation: N Engl J Med 2006;355(16):1660–1671. DOI: 10.1056/NEJMoa061752

Link: https://doi.org/10.1056/NEJMoa061752

Bottom Line

Carotid endarterectomy produced substantially lower 30-day and 6-month rates of stroke or death than carotid stenting in patients with symptomatic ≥60% carotid stenosis; the trial was stopped early for safety and futility, and stenting was not noninferior to endarterectomy.

        Major Points
        Multicenter, publicly funded, randomized noninferiority trial in 30 French centers (20 academic, 10 nonacademic); planned 872 patients but stopped after 527 for safety and futility (September 2005).
30-day any stroke or death: 3.9% with endarterectomy vs 9.6% with stenting (RR 2.5, 95% CI 1.2–5.1, P=0.01); noninferiority not met.
6-month any stroke or death: 6.1% vs 11.7% (P=0.02); results consistent across centers and physician experience strata.
Cranial-nerve injury was common after endarterectomy (7.7%) and rare after stenting (1.1%), but was largely mild and mostly resolved by 6 months.
Among stented patients, 30-day stroke or death was much lower with cerebral protection (7.9%) than without (25%, P=0.03), leading the safety committee to mandate protection devices in February 2003.
EVA-3S is a foundational trial (with SPACE, ICSS, CREST) informing guideline preference for carotid endarterectomy over stenting in most symptomatic patients, especially those ≥70 years.

      

Design

Study Type: Multicenter randomized controlled noninferiority trial

Randomization: 1

Blinding: Open-label treatment; outcome events adjudicated by a blinded events committee (except when local complications made blinding impossible)

Enrollment Period: November 2000 – September 2005 (stopped early)

Follow-up Duration: 30 days and 6 months reported; long-term follow-up ongoing at publication

Centers: 30

Countries: France

Sample Size: 527

Analysis: Intention-to-treat for 6-month outcomes; on-treatment (all randomized patients who underwent carotid repair) for 30-day outcomes

Inclusion Criteria

Age ≥18 years
Hemispheric or retinal transient ischemic attack, nondisabling stroke, or retinal infarct within 120 days before enrollment
Symptomatic ipsilateral carotid stenosis 60–99% by NASCET method (≥70% initially, lowered to ≥60% in October 2003)
Stenosis confirmed by catheter angiography or by both duplex ultrasonography and MR angiography
Considered a suitable candidate for both carotid endarterectomy and carotid stenting

Exclusion Criteria

Modified Rankin score ≥3 (disabling stroke)
Nonatherosclerotic carotid disease
Severe tandem lesions (proximal common-carotid or intracranial stenosis worse than the cervical lesion)
Previous revascularization of the symptomatic stenosis
Bleeding disorder; uncontrolled hypertension or diabetes; unstable angina
Contraindication to heparin, ticlopidine, or clopidogrel
Life expectancy <2 years
Percutaneous or surgical intervention within 30 days before or after the study procedure

Baseline Characteristics

Characteristic	Endarterectomy (N=259)	Stenting (N=261)
Mean Age (yr)	70.3 ± 10.7	69.1 ± 10.2
Age ≥75 yr	40.5%	32.2%
Male sex	78.0%	72.4%
Hypertension	72.6%	73.6%
Diabetes	25.5%	22.2%
Hypercholesterolemia	55.6%	57.9%
Current tobacco use	23.6%	24.1%
Systolic BP (mmHg)	140.8 ± 17.8	140.9 ± 16.5
BMI (kg/m²)	26.3 ± 4.1	26.1 ± 4.6
History of stroke	20.1%	12.6%
History of TIA	23.2%	25.3%
Prior MI	13.1%	10.7%
Peripheral arterial disease	11.6%	14.2%
Prior antiplatelet therapy	52.5%	49.0%
Prior lipid-lowering therapy	48.3%	49.4%
Qualifying event: ischemic stroke	53.7%	48.7%
Qualifying event: cerebral TIA	30.1%	36.4%
Stenosis 90–99%	40.9%	39.8%
Contralateral carotid occlusion	1.2%	5.0%
Median days randomization→treatment	6.0 (IQR 2–10)	6.0 (IQR 3–9)

Arms

Field	Control	Carotid-artery stenting
Intervention	Carotid endarterectomy per customary practice (patch 50.2%, no patch 20.6%, eversion 24.5%); general anesthesia 73%; median operative time 80 min; surgeons required to have performed ≥25 endarterectomies in the previous year.	Transfemoral carotid stenting with approved stents and cerebral-protection devices (mandatory after February 2003; used in 91.9% overall); dual antiplatelet therapy (aspirin 100–300 mg + clopidogrel 75 mg or ticlopidine 500 mg) 3 days before and 30 days after; interventionalists required to have performed ≥12 carotid stenting or ≥35 supraaortic stenting (of which ≥5 carotid) procedures, or to work under a tutor.
Duration	Single procedure with 30-day and 6-month follow-up	Single procedure with 30-day and 6-month follow-up

Outcomes

Outcome	Type	Control	HR / OR / RR	P-value
Any stroke or death within 30 days after treatment	Primary			0.01
30-day nonfatal stroke	Secondary		3.3	0.004
30-day death (any cause)	Secondary		0.7	0.68
30-day disabling stroke or death	Secondary		2.2	0.26
30-day TIA	Secondary		3	0.28
30-day myocardial infarction	Secondary		0.5	0.62
30-day bradycardia or hypotension requiring treatment	Secondary		not computable	<0.001
30-day cranial-nerve injury	Secondary		0.15	<0.001
30-day major local complications	Secondary		2.6	NS
30-day systemic complications (mainly pulmonary)	Secondary		0.6	0.42
Median hospital stay (days)	Secondary			0.01
30-day stroke or death + ipsilateral stroke by 6 mo	Secondary			0.008
30-day stroke or death + any stroke by 6 mo	Secondary			0.007
Any stroke or death within 6 months	Secondary			0.02
30-day stroke or death (primary safety endpoint)	Adverse	3.9% endarterectomy vs 9.6% stenting (RR 2.5, P=0.01)
Cranial-nerve injury	Adverse	7.7% endarterectomy vs 1.1% stenting (P<0.001) — largely mild/moderate; most resolved by 6 mo
Bradycardia/hypotension needing treatment	Adverse	0% vs 4.2% (P<0.001)
Major local complications (femoral pseudoaneurysm, hematoma, occlusion, infection)	Adverse	1.2% vs 3.1% (NS)
Systemic (mainly pulmonary) complications	Adverse	3.1% vs 1.9% (NS)

Subgroup Analysis

Relative risk of 30-day stroke or death (stenting vs endarterectomy) was consistent across enrollment-volume strata (<21 patients RR 1.9; 21–40 RR 3.3; >40 RR 2.7; P=0.83 for heterogeneity) and across levels of interventional-physician experience (experienced 10.5%, tutored during training 7.1%, tutored after training 12.3%; P=0.54). Age-adjusted RR 2.4 (1.2–4.8); stroke-history-adjusted RR 2.6 (1.3–5.2). Cerebral-protection use markedly reduced 30-day stroke or death after stenting (7.9% vs 25% without protection, P=0.03). Antiplatelet regimen (dual 9.0% vs single 11.1%) did not differ (P=0.75).

Criticisms

Stopped early for safety/futility at 527 of a planned 872 patients — early stopping can overestimate the treatment effect (‘random-high’ phenomenon).
Interventionalist experience threshold was low (≥12 carotid or ≥35 supraaortic-with-≥5-carotid stents) and some centers used tutors; a learning-curve effect on stenting outcomes cannot be excluded.
Cerebral-protection devices were only mandated in February 2003 (≈16 months into enrollment); early stenting patients treated without protection had markedly higher stroke rates.
Multiple stent and protection-device models were permitted — heterogeneity of technique may limit generalizability.
Open-label design with unblinded operators; only the events committee was blinded (and not for local complications).
Conducted entirely in France in academic and nonacademic centers — external validity to other health systems and to asymptomatic or high-surgical-risk populations is limited.
Endarterectomy 30-day event rate (3.9%) was lower than in the pivotal NASCET/ECST trials, which may widen the observed relative risk against stenting.

Funding

Publicly funded by the Programme Hospitalier de Recherche Clinique of the French Ministry of Health (AOM 97066), Assistance Publique–Hôpitaux de Paris. Two authors reported lecture fees from device manufacturers (ev3, Guidant, Cordis, Cook); no other conflicts declared.

Based on: EVA-3S (N Engl J Med, 2006)

Authors: Mas JL, Chatellier G, Beyssen B, et al. (EVA-3S Investigators)

Citation: N Engl J Med 2006;355(16):1660–1671. DOI: 10.1056/NEJMoa061752

Content summarized and formatted by NeuroTrials.ai.

EVA-3S

(2006)

Objective

Test whether carotid-artery stenting is noninferior to carotid endarterectomy for the risk of any stroke or death within 30 days in patients with symptomatic severe (≥60%) carotid stenosis.

Study Summary

• Trial stopped early for safety and futility after 527/872 planned patients.
• 30-day stroke or death: 9.6% with stenting vs 3.9% with endarterectomy (RR 2.5, 95% CI 1.2–5.1; P=0.01) — stenting was NOT noninferior.
• 6-month any stroke or death: 11.7% vs 6.1% (P=0.02); disabling stroke or death at 30d 3.4% vs 1.5% (RR 2.2, NS).
• Cranial-nerve injury lower with stenting (1.1% vs 7.7%, P<0.001); hospital stay 1 day shorter (median 3 vs 4 days, P=0.01).

Intervention

Carotid-artery stenting (transfemoral, with cerebral protection devices required after Feb 2003; dual antiplatelet 3 days before to 30 days after) versus carotid endarterectomy (customary technique, surgeons with ≥25 endarterectomies in prior year).

Inclusion Criteria

Age ≥18 yr; hemispheric or retinal TIA or nondisabling stroke (or retinal infarct) within 120 days; symptomatic ipsilateral carotid stenosis 60–99% by NASCET method confirmed by catheter angiography or duplex + MR angiography; suitable candidate for both procedures.

Study Design

Arms: Carotid endarterectomy (n=259 treated) vs carotid-artery stenting (n=261 treated); 527 randomized in 30 French centers.

Patients per Arm: Endarterectomy 259 / Stenting 261 (527 randomized, 520 underwent repair and analyzed for 30-day outcome)

Outcome

• Primary: 30-day any stroke or death — 3.9% endarterectomy vs 9.6% stenting; RR 2.5 (95% CI 1.2–5.1), P=0.01; noninferiority not met (95% CI of difference 2.1–9.3% exceeded 2% margin).
• 30-day nonfatal stroke 2.7% vs 8.8% (RR 3.3, 1.4–7.5, P=0.004); disabling stroke or death 1.5% vs 3.4% (RR 2.2, 0.7–7.2, P=0.26).
• 6-month any stroke or death 6.1% vs 11.7% (P=0.02); 30-day death or any stroke through 6 mo 4.6% vs 10.9% (P=0.007).
• Cranial-nerve injury 7.7% (endart.) vs 1.1% (stent), P<0.001; bradycardia/hypotension 0% vs 4.2%, P<0.001.
• Stenting-with-cerebral-protection 30-d stroke or death 7.9% vs stenting-alone 25% (P=0.03).

Clinical Question

In patients with symptomatic severe (≥60%) carotid stenosis, is carotid-artery stenting noninferior to carotid endarterectomy for the composite of any stroke or death within 30 days after treatment?

Bottom Line

Major Points

Multicenter, publicly funded, randomized noninferiority trial in 30 French centers (20 academic, 10 nonacademic); planned 872 patients but stopped after 527 for safety and futility (September 2005).
30-day any stroke or death: 3.9% with endarterectomy vs 9.6% with stenting (RR 2.5, 95% CI 1.2–5.1, P=0.01); noninferiority not met.
6-month any stroke or death: 6.1% vs 11.7% (P=0.02); results consistent across centers and physician experience strata.
Cranial-nerve injury was common after endarterectomy (7.7%) and rare after stenting (1.1%), but was largely mild and mostly resolved by 6 months.
Among stented patients, 30-day stroke or death was much lower with cerebral protection (7.9%) than without (25%, P=0.03), leading the safety committee to mandate protection devices in February 2003.
EVA-3S is a foundational trial (with SPACE, ICSS, CREST) informing guideline preference for carotid endarterectomy over stenting in most symptomatic patients, especially those ≥70 years.

Study Design

Study Type: Multicenter randomized controlled noninferiority trial
Randomization: Yes
Blinding: Open-label treatment; outcome events adjudicated by a blinded events committee (except when local complications made blinding impossible)
Sample Size: 527
Follow-up: 30 days and 6 months reported; long-term follow-up ongoing at publication
Centers: 30
Countries: France

Primary Outcome

Definition: Any stroke or death within 30 days after treatment

Control	Intervention	HR/OR	P-value
-	-	- (1.2–5.1)	0.01

Limitations & Criticisms

Stopped early for safety/futility at 527 of a planned 872 patients — early stopping can overestimate the treatment effect (‘random-high’ phenomenon).
Interventionalist experience threshold was low (≥12 carotid or ≥35 supraaortic-with-≥5-carotid stents) and some centers used tutors; a learning-curve effect on stenting outcomes cannot be excluded.
Cerebral-protection devices were only mandated in February 2003 (≈16 months into enrollment); early stenting patients treated without protection had markedly higher stroke rates.
Multiple stent and protection-device models were permitted — heterogeneity of technique may limit generalizability.
Open-label design with unblinded operators; only the events committee was blinded (and not for local complications).
Conducted entirely in France in academic and nonacademic centers — external validity to other health systems and to asymptomatic or high-surgical-risk populations is limited.
Endarterectomy 30-day event rate (3.9%) was lower than in the pivotal NASCET/ECST trials, which may widen the observed relative risk against stenting.

Citation

N Engl J Med 2006;355(16):1660–1671. DOI: 10.1056/NEJMoa061752

View Original Publication →

EVA-3S

Endarterectomy versus Stenting in Patients with Symptomatic Severe Carotid Stenosis

Bottom Line

Major Points

Design

Inclusion Criteria

Exclusion Criteria

Baseline Characteristics

Arms

Outcomes

Subgroup Analysis

Criticisms

Funding

Ask about EVA-3S