Skip to main content

Foundational Trials

MR CLEAN

Image: Shutterstock / ChaNaWiT

Publication

  • Title: A Randomized Trial of Intraarterial Treatment for Acute Ischemic Stroke
  • Acronym: MR CLEAN (Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischaemic Stroke in the Netherlands)
  • Year: 2015
  • Journal published in: New England Journal of Medicine
  • Citation: Berkhemer OA, Fransen PSS, Beumer D, et al; MR CLEAN Investigators. A randomized trial of intraarterial treatment for acute ischemic stroke. N Engl J Med. 2015;372:11-20.

Context & Rationale

  • Background
    • Proximal anterior-circulation large-vessel occlusion (LVO) causes a large proportion of disabling strokes and has poor natural history without early reperfusion.
    • Intravenous alteplase (when eligible) achieves low early recanalisation rates in proximal LVO; a substantial “reperfusion gap” remained for internal carotid artery terminus and M1 occlusions.
    • Earlier endovascular strategies (intra-arterial fibrinolysis and first-generation devices) and several pre-2015 RCTs were limited by delayed workflows, variable imaging confirmation of occlusion, and limited use of modern stent-retrievers.
    • By the early 2010s, stent-retrievers and better imaging/workflow created biological plausibility for benefit, but definitive pragmatic RCT evidence was still lacking in routine care settings.
  • Research Question/Hypothesis
    • In adults with acute ischaemic stroke due to proximal anterior-circulation occlusion who can be treated within 6 hours, does intraarterial treatment added to usual care improve 90‑day functional outcome compared with usual care alone?
    • Hypothesis: intraarterial treatment would shift the full modified Rankin Scale (mRS) distribution towards less disability at 90 days.
  • Why This Matters
    • Functional outcome after LVO stroke is the dominant driver of long-term dependence, institutional care, and indirect societal cost.
    • A positive pragmatic multicentre RCT would change acute stroke pathways (prehospital triage, inter-hospital transfer, round-the-clock endovascular capability).
    • For critical care teams, improved reperfusion impacts neuro-ICU case-mix (malignant infarction, ventilatory requirements, haemodynamic targets, sedation strategy) and downstream rehabilitation demand.

Design & Methods

  • Research Question: Among patients with acute ischaemic stroke from confirmed proximal anterior-circulation occlusion treatable within 6 hours, does intraarterial treatment plus usual care improve 90‑day disability (mRS shift) versus usual care alone?
  • Study Type: Randomised, multicentre (16 centres), investigator-initiated, pragmatic superiority trial; open-label treatment with blinded endpoint assessment (PROBE design); acute stroke pathway (emergency/CT/CTA with transfer to endovascular-capable centre if required).
  • Population:
    • Setting: Acute stroke care in the Netherlands; enrolment required vessel imaging confirmation of proximal anterior-circulation occlusion.
    • Key inclusion criteria: Age ≥18; NIHSS ≥2; non-contrast CT excluding intracranial haemorrhage; proximal anterior-circulation occlusion (ICA terminus/M1/M2/A1–A2) on CTA/MRA/angiography; eligibility for intraarterial treatment within 6 hours of symptom onset.
    • Key exclusion criteria: Standard contraindications to IV thrombolysis (when considered), including persistent blood pressure >185/110 mmHg despite treatment; blood glucose <2.7 or >22.2 mmol/L; other protocol-defined exclusions (e.g., clinical/imaging features precluding endovascular therapy).
  • Intervention:
    • Tested strategy: Intraarterial treatment (arterial catheterisation with mechanical thrombectomy, intra-arterial thrombolytic therapy, or both) in addition to usual care.
    • Time target: Initiation within 6 hours of symptom onset; median onset-to-groin puncture 260 minutes (IQR 210–313) in the intervention group.
    • Delivered treatment (as-treated descriptors): Intraarterial treatment initiated in 196/233 (84.1%); retrievable stents used in 190/233 (81.5%); additional intra-arterial thrombolytics in 24/233 (10.3%); acute carotid artery stenting in 30/233 (12.9%); general anaesthesia in 88/233 (37.8%).
  • Comparison:
    • Control strategy: Usual care alone, including IV alteplase when eligible and other standard acute stroke management; no protocolised endovascular “rescue” outside trial allocation.
    • Co-interventions: IV alteplase use was high in both groups (87.1% intervention vs 90.6% control).
  • Blinding: Open-label delivery of endovascular therapy; functional outcomes assessed by personnel blinded to treatment assignment (primary endpoint mRS at 90 days).
  • Statistics: Power calculation: 500 patients to provide 82% power (β=0.18) at α=0.05 to detect a 10% absolute increase in the cumulative proportion of patients with mRS 0–3 at 90 days, assuming 10% cross-over2; primary analysis by intention-to-treat using ordinal logistic regression (common odds ratio) adjusted for prespecified prognostic variables.
  • Follow-Up Period: 90 days for primary disability outcome; key secondary outcomes included NIHSS at 24 hours and at 5–7 days/discharge, and imaging outcomes at follow-up CTA and 5–7 day CT.

Key Results

This trial was not stopped early. The target sample size (500 analysed participants) was reached.

Outcome Intraarterial treatment + usual care Usual care alone Effect p value / 95% CI Notes
Primary: mRS at 90 days (shift analysis) Median 4 Median 5 Adjusted common OR 1.67 95% CI 1.21 to 2.30 Ordinal logistic regression; common OR >1 favours intervention
Functional independence (mRS 0–2) at 90 days 76/233 (32.6%) 51/267 (19.1%) Adjusted OR 2.16 95% CI 1.39 to 3.38 Absolute difference 13.5 percentage points (95% CI 5.9 to 21.2)
mRS 0–3 at 90 days 119/233 (51.1%) 95/267 (35.6%) Adjusted OR 2.03 95% CI 1.36 to 3.03 Absolute difference 15.5 percentage points (95% CI 5.9 to 25.1)
Barthel Index 19–20 at 90 days 99/215 (46.0%) 73/245 (29.8%) Adjusted OR 2.11 95% CI 1.42 to 3.14 Non-complete data for this endpoint (denominators < randomised N)
NIHSS at 24 hours Median 13 Median 16 Adjusted β 2.3 95% CI 1.0 to 3.5 β represents lower NIHSS (better) in the intervention group
NIHSS at 5–7 days or discharge Median 8 Median 14 Adjusted β 2.9 95% CI 1.5 to 4.3 Direction consistent with improved early neurological recovery
No intracranial occlusion on follow-up CTA 141/187 (75.4%) 68/207 (32.9%) Adjusted OR 6.88 95% CI 4.44 to 10.66 Imaging endpoint with incomplete ascertainment (CTA not in all)
Infarct volume at 5–7 days (mL) Median 49 Median 79 Adjusted β 19 95% CI 3 to 34 Smaller infarct volume in intervention group; incomplete data
Symptomatic intracranial haemorrhage 18/233 (7.7%) 17/267 (6.4%) RR 1.21 95% CI 0.60 to 2.43; P=0.58 No statistically significant excess; wide CI
Any intracranial haemorrhage 68/233 (29.2%) 49/267 (18.4%) RR 1.59 95% CI 1.08 to 2.35; P=0.02 Imaging-based; clinical relevance varies by type/location
New ischaemic stroke in different vascular territory 13/233 (5.6%) 1/267 (0.4%) RR 14.91 95% CI 1.97 to 112.68; P<0.001 Important procedural harm signal; imprecise CI due to low counts
Death within 30 days 44/233 (18.9%) 49/267 (18.4%) RR 1.03 95% CI 0.71 to 1.49; P=0.88 Trial reported no significant mortality difference at 7, 30, or 90 days
  • Disability improved across the full mRS distribution: adjusted common OR 1.67 (95% CI 1.21 to 2.30), with median mRS 4 vs 5.
  • Clinically interpretable gains were present for independence: mRS 0–2 was 32.6% vs 19.1% (adjusted OR 2.16; 95% CI 1.39 to 3.38; absolute difference 13.5 percentage points; 95% CI 5.9 to 21.2).
  • Biological separation was strong on follow-up CTA: no intracranial occlusion 75.4% vs 32.9% (adjusted OR 6.88; 95% CI 4.44 to 10.66), with a key procedural harm signal of new-territory infarction (5.6% vs 0.4%; RR 14.91; 95% CI 1.97 to 112.68).
  • Prespecified subgroup analyses (primary outcome; adjusted common OR, 95% CI):
    • Age ≥80: 3.24 (1.22 to 8.62) vs age <80: 1.60 (1.13 to 2.28).
    • NIHSS ≥20: 1.85 (1.06 to 3.21) (point estimate suggests retained benefit in severe stroke).
    • ICA terminus occlusion present: 2.43 (1.24 to 4.77) vs absent: 1.61 (1.11 to 2.33).
    • Onset-to-randomisation ≥120 minutes: 1.69 (1.21 to 2.38) (early <120 minutes underpowered: 1.57; 0.51 to 4.85).

Internal Validity

  • Randomisation and Allocation:
    • Central web-based randomisation; stratified for centre, baseline NIHSS (≤14 vs >14), planned IV alteplase use, and planned endovascular modality (mechanical vs intra-arterial thrombolysis) minimised predictable allocation and centre-level confounding2.
    • Group sizes were imbalanced (233 vs 267), but baseline characteristics were broadly similar; key observed imbalances (e.g., left-hemisphere strokes and IV alteplase use) favoured the control group.
  • Drop out or exclusions:
    • 502 randomised; 2 withdrew consent immediately after randomisation and were excluded, leaving 500 analysed participants (a minor deviation from strict ITT, but numerically trivial).
    • Primary endpoint ascertainment at 90 days was effectively complete for the analysed cohort (no signal of differential loss affecting mRS).
  • Performance/Detection Bias:
    • Treatment could not be blinded; co-interventions and post-procedure care could, in theory, differ by allocation.
    • Primary outcome assessment was blinded, reducing detection bias for mRS; many secondary outcomes were objective (CTA recanalisation, infarct volume) but had incomplete ascertainment.
  • Protocol Adherence:
    • Intraarterial treatment was initiated in 196/233 (84.1%); 37/233 (15.9%) did not receive intraarterial therapy (e.g., no occlusion at angiography [n=21], clinical improvement [n=3], access failure [n=3], no clot retrieved [n=5]).
    • Device and technique heterogeneity: retrievable stents in 190/233 (81.5%); additional intra-arterial thrombolytics in 24/233 (10.3%); acute carotid stenting in 30/233 (12.9%).
  • Baseline Characteristics:
    • Illness severity and timing were comparable: age 65.8 vs 65.7 years (median); NIHSS 17 vs 18 (median); onset-to-randomisation 204 vs 196 minutes (median).
    • High alteplase exposure in both arms (87.1% vs 90.6%) supports contemporaneous “best medical therapy” as baseline care.
    • Occlusion distribution typical of proximal LVO: M1 66.1% vs 62.0%; ICA terminus 23.6% vs 28.2%; extracranial ICA occlusion 32.2% vs 26.3%.
  • Heterogeneity:
    • Pragmatic inclusion and variable device/anaesthesia choices increase heterogeneity; this generally biases towards the null, supporting robustness of observed benefit.
    • Prespecified subgroup effects were directionally consistent and did not identify a subgroup with clear evidence of harm.
  • Timing:
    • Median onset-to-groin puncture was 260 minutes (IQR 210–313), consistent with “early window” thrombectomy and congruent with the trial hypothesis.
    • Workflow delays remain clinically relevant because time-to-reperfusion is a major determinant of infarct growth; MR CLEAN’s modest median times likely diluted achievable benefit compared with later, faster systems.
  • Dose:
    • Angiographic “dose” (successful reperfusion) was moderate: mTICI 2b/3 achieved in 115/196 (58.7%).
    • General anaesthesia occurred in 88/233 (37.8%), potentially influencing haemodynamics and workflow; the trial was not designed to isolate anaesthetic strategy effects.
  • Separation of the Variable of Interest:
    • Objective separation demonstrated on follow-up CTA: no intracranial occlusion 75.4% (141/187) vs 32.9% (68/207); adjusted OR 6.88 (95% CI 4.44 to 10.66).
    • Neurological recovery separation: NIHSS at 24 hours median 13 vs 16; adjusted β 2.3 (95% CI 1.0 to 3.5); NIHSS at 5–7 days/discharge median 8 vs 14; adjusted β 2.9 (95% CI 1.5 to 4.3).
  • Key Delivery Aspects:
    • Imaging-confirmed occlusion prior to randomisation is a major design strength compared with earlier neutral endovascular RCTs.
    • Non-treatment within the intervention arm (15.9%) and only moderate reperfusion rates imply that “protocol purity” and technical performance could plausibly increase effect size in later practice.
  • Crossover:
    • Crossover occurred in 18/500 (3.6%): 8 allocated to intervention crossed to control and 10 allocated to control crossed to intervention (low risk of biasing effect estimates).
  • Adjunctive therapy use:
    • IV alteplase use was slightly higher in control (90.6% vs 87.1%), which would tend to attenuate (not exaggerate) the observed benefit of the intraarterial strategy.
  • Outcome Assessment:
    • Primary endpoint (mRS at 90 days) is clinically meaningful and patient-centred; blinded assessment supports reliability.
    • Imaging endpoints (CTA patency and infarct volume) strengthen mechanistic plausibility but were not complete for all participants.
  • Statistical Rigor:
    • Primary analysis matched the ordinal nature of mRS (shift analysis) with prespecified adjustment, improving efficiency and reducing residual confounding.
    • Sensitivity adjustment replacing “previous stroke” with prestroke mRS yielded similar results (adjusted common OR 1.62; 95% CI 1.18 to 2.23), supporting robustness.

Conclusion on Internal Validity: Overall, internal validity appears strong to moderate: randomisation and blinded endpoint assessment support causal inference, and objective mechanistic endpoints demonstrate separation; key limitations are open-label care, incomplete delivery of intraarterial therapy in 15.9% of the intervention arm, and incomplete imaging ascertainment for some secondary outcomes.

External Validity

  • Population Representativeness:
    • Broad pragmatic enrolment (age range 23–96; median NIHSS 17–18) and high alteplase uptake reflect routine early-window LVO care in a mature stroke system.
    • Confirmed proximal occlusion (CTA/MRA/angiography) enhances applicability to contemporary stroke networks that rely on vascular imaging for triage.
  • Applicability:
    • Most generalisable to anterior-circulation proximal LVO within 6 hours, managed in systems with rapid imaging, transfer logistics, and 24/7 endovascular capability.
    • Less applicable to: posterior circulation occlusions, distal occlusions beyond inclusion, very late/unknown onset, and settings without consistent access to neurointerventional teams and follow-up imaging.
    • Technique and workflow have evolved since MR CLEAN (higher reperfusion rates and faster door-to-reperfusion in many centres), so absolute outcomes may differ while direction of effect remains relevant.

Conclusion on External Validity: Generalisability is good for early-window anterior LVO in well-resourced stroke systems, but is limited in settings lacking rapid vascular imaging/endovascular access or in patient groups outside the trial’s anatomic and temporal eligibility.

Strengths & Limitations

  • Strengths:
    • First large pragmatic multicentre RCT demonstrating clear functional benefit of an intraarterial strategy for confirmed proximal anterior-circulation occlusion.
    • Clinically meaningful primary endpoint (mRS shift) with blinded assessment; high follow-up completeness for primary outcome.
    • Mechanistic coherence: marked improvement in vessel patency on follow-up CTA and reduced infarct volume.
    • High background rate of alteplase use supports “add-on” interpretation (thrombectomy on top of modern medical therapy).
  • Limitations:
    • Open-label treatment introduces potential performance bias in supportive care and rehabilitation intensity (mitigated by blinded mRS assessment).
    • Incomplete protocol delivery: 15.9% of the intervention arm did not receive intraarterial therapy; reperfusion rate (mTICI 2b/3) was 58.7%—lower than in later-generation practice.
    • Device/technique and anaesthesia heterogeneity complicate attribution to any single procedural approach and may increase between-centre variability.
    • Important procedural harm signal: more new-territory infarcts (5.6% vs 0.4%) and higher “any intracranial haemorrhage” (29.2% vs 18.4%).

Interpretation & Why It Matters

  • Clinical practice
    • MR CLEAN establishes intraarterial treatment (predominantly mechanical thrombectomy) as an effective strategy to improve disability outcomes in early-window anterior-circulation LVO when added to best medical therapy.
    • The magnitude of functional improvement (shift in mRS distribution; 13.5% absolute increase in mRS 0–2) is clinically substantial at a population level and translates into fewer patients requiring long-term institutional care.
  • Mechanistic coherence
    • The large difference in follow-up CTA patency (75.4% vs 32.9%) supports a reperfusion-mediated causal pathway rather than a chance clinical signal.
    • Reduction in infarct volume (median 49 vs 79 mL) and improved early NIHSS corroborate downstream biological benefit.
  • Health-system implications
    • Benefit depends on rapid identification and transfer of eligible patients; system-level metrics (door-to-imaging, imaging-to-groin puncture, reperfusion times) become quality targets.
    • For critical care services, effective thrombectomy shifts the burden from malignant infarction management to earlier extubation, rehabilitation planning, and prevention of complications in recovering patients.

Controversies & Subsequent Evidence

  • “Why did MR CLEAN succeed when earlier endovascular RCTs were neutral?”
    • Key differentiator: mandatory vessel imaging confirmation of proximal occlusion before randomisation, aligning the intervention with the biological target population.
    • Contemporary devices and techniques were used frequently (retrievable stents 81.5%), despite overall reperfusion rates being modest compared with later trials.
  • Editorial perspective: system performance and technique were central to translating efficacy into effectiveness
    • The accompanying editorial emphasised that benefit depends on rapid workflow and experienced centres, and framed MR CLEAN as the beginning of a new era requiring disciplined systems of care1.
  • Procedural risk trade-offs
    • New-territory infarction occurred more often with intraarterial treatment (5.6% vs 0.4%; RR 14.91), underscoring that net benefit requires procedural expertise and attention to embolic protection/technique.
    • Symptomatic intracranial haemorrhage was similar (7.7% vs 6.4%), supporting acceptable haemorrhagic risk in an alteplase-rich population.
  • Rapid confirmatory RCTs and end of equipoise
    • Multiple 2015 trials in similar populations (e.g., ESCAPE) corroborated improved functional outcomes with thrombectomy, accelerating guideline adoption3.
    • Patient-level meta-analysis (HERMES) confirmed consistent benefit across a broad range of ages and severity, and strengthened the evidentiary base for system-wide implementation4.
  • Extension beyond the 6-hour window
    • Later trials using advanced imaging selection expanded thrombectomy benefit into later time windows (DAWN, DEFUSE 3), reshaping triage and transfer paradigms for late-presenting LVO56.
  • Guideline incorporation
    • Major stroke guidelines incorporated thrombectomy for eligible anterior-circulation LVO and emphasised speed and imaging-based selection (AHA/ASA 2018 guideline and 2019 update)78.

Summary

  • MR CLEAN demonstrated that intraarterial treatment added to usual care improves 90-day disability outcomes after confirmed proximal anterior-circulation occlusion (adjusted common OR 1.67; 95% CI 1.21 to 2.30).
  • Functional independence (mRS 0–2) increased from 19.1% to 32.6% (adjusted OR 2.16; 95% CI 1.39 to 3.38; absolute difference 13.5 percentage points; 95% CI 5.9 to 21.2).
  • Mechanistic endpoints aligned: follow-up CTA patency was 75.4% vs 32.9% (adjusted OR 6.88; 95% CI 4.44 to 10.66), and infarct volume was smaller (median 49 vs 79 mL).
  • Key harms: more any intracranial haemorrhage (29.2% vs 18.4%) and more new-territory strokes (5.6% vs 0.4%), while symptomatic intracranial haemorrhage was similar (7.7% vs 6.4%).
  • The trial’s pragmatic design and imaging-confirmed eligibility made it practice-changing and set the evidentiary foundation for modern thrombectomy pathways.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Only references with DOI identifiers available from the provided materials or high-certainty canonical DOIs are listed here; additional high-quality post-2019 guidelines, registries, and meta-analyses exist but are not enumerated in this page.

Overall Takeaway

MR CLEAN is “landmark” because it was the first large pragmatic RCT to show that an intraarterial endovascular strategy, delivered in routine multicentre practice and targeted to imaging-confirmed proximal LVO, improves patient-centred disability outcomes at 90 days. Its mechanistic coherence (better patency, smaller infarcts) and clinically meaningful functional gains catalysed rapid confirmation by subsequent trials and fundamentally reshaped modern stroke systems of care.

Overall Summary

  • Intraarterial treatment improved 90‑day disability (adjusted common OR 1.67) and increased functional independence (32.6% vs 19.1%).
  • Strong mechanistic separation on CTA (75.4% vs 32.9% no intracranial occlusion) supports a reperfusion-mediated effect.
  • Net benefit coexisted with procedural harms (notably new-territory stroke), reinforcing the importance of expertise and systems performance.

Bibliography