PDF: IMS III
(2013)Objective
In patients with moderate-to-severe acute ischemic stroke treated with IV t-PA within 3 hours of onset, does adding endovascular therapy improve 90-day functional independence versus IV t-PA alone?
Study Summary
• No benefit in either NIHSS stratum: score ≥20 difference +6.8 pp (95% CI −4.4 to 18.1); score 8–19 difference −1.0 pp (95% CI −10.8 to 8.8).
• Safety similar: 90-day mortality 19.1% vs 21.6% (P=0.52); sICH within 30h 6.2% vs 5.9% (P=0.83); asymptomatic ICH higher with endovascular (27.4% vs 18.9%, P=0.01).
Intervention
IV t-PA (0.9 mg/kg over 1 h) followed by endovascular therapy (intra-arterial t-PA via standard microcatheter or EKOS MicroSonic system, or thrombectomy with Merci, Penumbra, or Solitaire FR) vs standard IV t-PA alone.
Inclusion Criteria
Age 18–82; IV t-PA started within 3 h of symptom onset; NIHSS ≥10 (or 8–9 with CTA-confirmed M1, ICA, or basilar occlusion after amendment 3); randomized within 40 min of t-PA initiation; angiographic procedure to begin ≤5 h and end ≤7 h from onset.
Study Design
Arms: Endovascular therapy after IV t-PA (n=434) vs IV t-PA alone (n=222); 2:1 randomization.
Patients per Arm: 434 vs 222 (N=656 randomized; planned 900, stopped early for futility; ITT analysis).
Outcome
• Subgroups: NIHSS ≥20 → +6.8 pp (95% CI −4.4 to 18.1); NIHSS 8–19 → −1.0 pp (95% CI −10.8 to 8.8); no significant interaction by time-to-tPA, age, sex, atrial fibrillation, or CTA-confirmed occlusion.
• Full mRS distribution non-significant overall (P=0.25 by van Elteren).
• Safety: 90-d mortality 19.1% (83/434) vs 21.6% (48/222), P=0.52; sICH ≤30 h 6.2% (27/434) vs 5.9% (13/222), P=0.83; parenchymal hematoma type 2 6.0% vs 6.3% (P=0.90); asymptomatic ICH 27.4% vs 18.9% (P=0.01); SAH 11.5% vs 5.8% (P=0.02); recurrent stroke at 90 d 5.1% vs 6.3% (P=0.54); device/procedural complication 16.1% in endovascular arm.
• Procedural reperfusion (endovascular, TICI 2–3): ICA 65%, M1 81%, M2 single 70%, M2 multiple 77%; TICI 2b–3: ICA 38%, M1 44%, M2 single 44%, M2 multiple 23%.
Bottom Line
Adding endovascular therapy after IV t-PA did not improve 90-day functional independence over IV t-PA alone (40.8% vs 38.7%; adjusted difference 1.5 pp, 95% CI −6.1 to 9.1) and was stopped early for futility. Safety was comparable, though asymptomatic ICH and SAH were more frequent with endovascular therapy. Because the trial used mostly first-generation devices (Merci/Penumbra) with prolonged time-to-groin puncture and did not routinely require CTA-confirmed LVO, it did not settle endovascular efficacy — later stent-retriever era trials (MR CLEAN, ESCAPE, EXTEND-IA, SWIFT PRIME, REVASCAT) did.
Major Points
- Phase 3, international, open-label, blinded-endpoint RCT at 58 centers in USA, Canada, Australia, and Europe (2006–2012); 2:1 randomization of IV t-PA + endovascular therapy vs IV t-PA alone.
- Stopped early for futility after 656 of planned 900 patients randomized (crossed pre-specified futility boundary on conditional power <20%).
- Primary outcome (mRS 0–2 at 90 d): 40.8% endovascular vs 38.7% IV t-PA alone; adjusted absolute difference 1.5 pp (95% CI −6.1 to 9.1); van Elteren P=0.25 for full mRS distribution.
- No benefit in either NIHSS stratum (≥20 or 8–19) and no significant interaction with age, sex, atrial fibrillation, time-to-tPA, or CTA-documented occlusion.
- 90-day mortality similar (19.1% vs 21.6%, P=0.52); symptomatic ICH within 30 h similar (6.2% vs 5.9%, P=0.83); parenchymal hematoma type 2 similar (6.0% vs 6.3%).
- Asymptomatic ICH (27.4% vs 18.9%, P=0.01) and subarachnoid hemorrhage (11.5% vs 5.8%, P=0.02) higher with endovascular therapy; device/procedural complications 16.1% in endovascular arm.
- Endovascular reperfusion (TICI 2–3) high: 65% ICA, 81% M1, 70% single M2, 77% multiple M2; TICI 2b–3 lower: 38% ICA, 44% M1, 44% single M2, 23% multiple M2. Better mRS with higher TICI (12.7% at TICI 0 → 71.4% at TICI 3, P<0.001).
- Devices reflected era: mostly intra-arterial t-PA, Merci, and Penumbra; only 4 patients received Solitaire (stent retriever); mean time from t-PA to groin puncture ~32 min longer than in IMS I.
- Routine CTA to require LVO was not mandatory throughout enrollment; roughly one-third of endovascular patients had no treatable occlusion at angiography, diluting a potential benefit.
Study Design
- Study Type
- International, phase 3, randomized, open-label, blinded-endpoint (PROBE) trial
- Randomization
- Yes
- Blinding
- Open-label treatment; 90-day mRS assessed by investigators unaware of treatment assignment (blinded endpoint).
- Sample Size
- 656
- Follow-up
- 90 days (primary outcome)
- Centers
- 58
- Countries
- USA, Canada, Australia, Netherlands, France, Switzerland, Germany, Spain
Primary Outcome
Definition: Modified Rankin Scale score ≤2 (functional independence) at 90 days, adjusted for baseline NIHSS stratum (8–19 vs ≥20); intention-to-treat with unfavorable imputation for missing/out-of-window assessments.
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| 38.7% (IV t-PA alone, n=222) | 40.8% (endovascular + IV t-PA, n=434) | - (−6.1 to 9.1) | Not significant (van Elteren P=0.25 for full mRS distribution) |
Limitations & Criticisms
- First-generation devices dominated: only 4 patients received Solitaire (stent retriever) before trial stopped; results predate the stent-retriever era demonstrated by MR CLEAN, ESCAPE, EXTEND-IA, SWIFT PRIME, and REVASCAT (2015).
- CTA-confirmed large-vessel occlusion was not required for the majority of enrollment — ~one-third of endovascular patients had no treatable occlusion at angiography, diluting a potential treatment effect.
- Time to endovascular treatment was ~32 minutes longer than in IMS I, undermining the reperfusion-efficacy relationship known to be time-dependent.
- Two-thirds-dose IV t-PA was used in the endovascular arm before amendment 5 (mid-2011); dose asymmetry across the trial complicates interpretation.
- Stopped early for futility after 656 of 900 patients — conditional-power stopping may under-detect a modest benefit, particularly in LVO subgroups.
- Open-label design with blinded endpoint (PROBE); acute imaging and management differences could bias assessment.
- Predominantly white population (only 11.8% Black, 2.5% Hispanic in endovascular arm) and upper age cap of 82 years limit generalizability.
- Endovascular arm used heterogeneous techniques (IA t-PA, EKOS, Merci, Penumbra, few stent retrievers), preventing device-specific inference.
- Higher rate of asymptomatic ICH and SAH in the endovascular arm was not clinically consequential here but signals procedural bleeding risk.
Citation
N Engl J Med 2013;368(10):893–903. DOI: 10.1056/NEJMoa1214300