PDF: JAM
(2014)Objective
Does bilateral direct extracranial–intracranial (EC-IC) bypass surgery reduce the incidence of rebleeding and adverse neurological events compared to conservative medical management in adults with hemorrhagic moyamoya disease?
Study Summary
Intervention
Multicenter prospective RCT; 22 centers in Japan; 80 patients with hemorrhagic moyamoya disease randomized to bilateral direct STA-MCA bypass (n=42) or conservative management (n=38); 5-year follow-up.
Inclusion Criteria
Age 16–65; mRS 0–2; ICH/IVH/SAH within preceding 12 months; ≥1 month post-stroke; bilateral angiographic moyamoya (Japanese MHW criteria); no prior EC-IC bypass
Study Design
Arms: Surgical: bilateral direct EC-IC bypass (STA-MCA anastomosis, n=42) | Nonsurgical: conservative medical management (n=38)
Patients per Arm: 42 vs 38
Outcome
• Secondary (rebleeding >3 months post-enrollment): 11.9% vs 31.6%; 2.7%/y vs 7.6%/y; P=0.042; HR 0.355 (95% CI 0.125–1.009)
• Perioperative complications: 8/84 procedures (9.5%); 7/8 transient; 0 permanent severe disability
Bottom Line
Bilateral direct EC-IC bypass significantly reduced the annualized rate of both all adverse events (3.2%/year vs 8.2%/year; P=0.048) and rebleeding attacks (2.7%/year vs 7.6%/year; P=0.042) in adults with hemorrhagic moyamoya disease, although Cox regression hazard ratios were statistically marginal with confidence intervals slightly crossing 1.0.
Major Points
- JAM was the first prospective randomized controlled trial for hemorrhagic moyamoya disease, planned in 1999, enrolled 2001–2008 across 22 expert Japanese neurosurgical centers, with final follow-up completed June 2013.
- Primary end point (any adverse event causing significant morbidity, mRS ≥3) occurred in 6/42 (14.3%) surgical vs 13/38 (34.2%) nonsurgical patients; annualized event rate 3.2%/year vs 8.2%/year (P=0.048 by log-rank); Cox HR 0.391 (95% CI 0.148–1.029, P=0.057).
- Secondary end point (rebleeding occurring >3 months after enrollment or related death/severe disability) occurred in 5/42 (11.9%) surgical vs 12/38 (31.6%) nonsurgical patients; annualized rate 2.7%/year vs 7.6%/year (P=0.042 by log-rank); Cox HR 0.355 (95% CI 0.125–1.009, P=0.052).
- Results were statistically marginal: log-rank P-values were just below 0.05, but Cox regression 95% CI upper limits slightly exceeded 1.0 for both endpoints (1.029 primary, 1.009 secondary), preventing definitive conclusions of surgical superiority.
- The originally calculated sample size was 160 patients (80/group) for 80% power to detect a difference between 8%/year (control) and 4%/year (surgical); reduced to 80 total in January 2006 because only 13.2 patients/year were being enrolled and completing recruitment by 2018 was deemed infeasible.
- Perioperative complications occurred in 8/84 surgical procedures (9.5%): local hyperperfusion (3 cases), TIA (1), seizure (1), local vasogenic edema (1), scalp bedsore (1), subcutaneous drain tube tear (1). Seven of 8 were clinically transient with no residual sequelae; 1 hyperperfusion patient had transient mRS worsening not meeting end-point criteria.
- Stratified randomization by bleeding type: Type A (anterior collateral hemorrhage — caudate/putamen) vs Type B (posterior collateral hemorrhage — thalamus/lateral ventricle trigone). Type A: 24 surgical, 21 nonsurgical; Type B: 18 surgical, 17 nonsurgical.
- Historical context: prior observational data (Kobayashi et al) showed 33.3% rebleeding over mean 6.7 years follow-up, estimating 7.09%/year annual rebleeding rate in conservatively managed patients — consistent with JAM control arm (7.6%/year).
- Surgical intervention required bilateral STA-MCA direct anastomosis; indirect bypass could supplement but not replace direct bypass; high-flow bypasses (venous graft, radial artery graft) were prohibited. Each side operated within 3 months of enrollment.
- One patient in the surgical group was lost to follow-up at 1.95 years due to murder (unrelated to medical condition); included as dropout in survival analysis, not as an end-point event. Mean follow-up: 4.32 years overall (4.46 years surgical, 4.17 years nonsurgical).
Study Design
- Study Type
- Multicenter, prospective, randomized, controlled trial
- Randomization
- Yes
- Blinding
- Unblinded (open-label) — surgery vs conservative care cannot be blinded; endpoint adjudication by executive and steering committee (neurologists/neurosurgeons) not blinded to allocation
- Sample Size
- 80
- Follow-up
- 5 years per patient; last patient completed June 2013
- Centers
- 22
- Countries
- Japan
Primary Outcome
Definition: Composite of all adverse events causing significant morbidity (mRS ≥3): (1) recurrent intracranial bleeding; (2) completed ischemic stroke causing significant morbidity; (3) significant morbidity or mortality from other medical cause; (4) requirement for EC-IC bypass in a nonsurgical patient due to progressive ischemic stroke or crescendo TIAs as determined by a registered neurologist. Asymptomatic bleeding detected incidentally on MRI was NOT counted.
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| - | - | - (0.148–1.029) | - |
Limitations & Criticisms
- Statistically marginal results: log-rank P-values were 0.048 and 0.042 (just below 0.05), while Cox regression 95% CI upper limits crossed 1.0 (1.029 for primary, 1.009 for secondary endpoint) — precluding definitive conclusions of surgical superiority.
- Small sample size (n=80 vs planned n=160): original sample size calculation assumed 8%/year event rate in control and 4%/year in surgical; with 80 patients, 80% power was achievable only if surgical event rate was <2.8%/year (which it was: 2.7%/year) — the trial was underpowered for its planned effect size.
- Unblinded design: neither patients, surgeons, nor outcome adjudicators were blinded to allocation, introducing potential assessment bias. Endpoint adjudication committee was also not blinded.
- Restricted generalizability: all 22 participating centers had documented expertise in moyamoya bypass surgery, and only registered surgeons could operate. Results may not apply to centers with less experience.
- Only 5-year outcomes reported; recurrent bleeding in moyamoya disease is known to occur beyond 10 years after initial hemorrhage, and long-term benefits (or risks) of bypass remain unknown.
- Exclusively Japanese population: moyamoya disease genetic background (RNF213 polymorphisms) and vascular anatomy may differ from other ethnic groups, limiting applicability to non-Japanese patients.
- Randomization was stratified by bleeding site (Type A vs B), but no powered subgroup analysis was performed to determine whether benefit was differential between these groups.
- One patient in the surgical group was lost to follow-up due to murder — handled as dropout; if instead treated as an event (worst-case analysis), results might differ.
Citation
Miyamoto S, et al. Stroke. 2014;45:1415–1421. DOI: 10.1161/STROKEAHA.113.004386