Long-Term Follow-Up of Extracranial-Intracranial Bypass Surgery for Symptomatic Cerebral Artery Occlusion

Yan Ma, MD1; Tao Wang, MD1; Wei Ni, MD2; et al

2025년 5월 22일

JAMA. Published online May 22, 2025. (Research Letter)

DOI: 10.1001/jama.2025.7225

Previous trials have shown no significant benefit of extracranial-intracranial (EC-IC) bypass surgery vs medical therapy in symptomatic patients with hemodynamically compromised internal carotid artery or middle carotid artery occlusion. CMOSS (Carotid and Middle Cerebral Artery Occlusion Surgery Study) was a multicenter, randomized clinical trial that compared EC-IC bypass surgery plus medical treatment vs medical treatment alone. The primary outcome was neutral at the 2-year follow up; the present report includes the longer-term follow-up results.

Methods

After finishing the initial preplanned follow-up, the study protocol was amended and approved for long-term follow-up of up to 10 years by institutional research ethics board at each center (Supplement 1). An addendum consent was required. Among 330 patients initially randomized, 324 patients were eligible and followed up through June 12, 2024. The primary outcome was a composite of any stroke or death within 30 days or ischemic stroke ipsilateral to the occluded artery beyond 30 days after randomization. Analyses were conducted in all eligible randomized participants and also in a per-protocol set; secondary outcomes and sensitivity analyses to address missing data were performed (eMethods in Supplement 2). The difference in the primary outcome was tested using a log-rank test, with the center as a stratification factor. P < .05 was considered statistically significant. χ2 or Fisher exact tests were used for categorical variables, t tests or Wilcoxon rank tests were used for continuous variables, and Kaplan-Meier analysis was used for the risk of time-to-event outcomes. All statistical tests were 2-sided and completed using SAS, version 9.4 (SAS Institute).

Results

Of 324 patients, 161 were assigned to the surgical group and 163 to the medical management alone group; the median (IQR) duration of follow-up was 7.6 (2.3-9.2) years. In the surgical and medical groups, respectively, 151 and 155 patients completed 2-year follow up, 24 and 22 were lost to long-term follow-up, and 2 and 3 withdrew consent, resulting in 255 patients (78.7%) with complete follow-up data. The primary outcome occurred in 18 of 161 patients (11.2%) in the surgical group and 32 of 163 (19.6%) in the medical group, which was a significant difference (relative risk [RR], 0.57 [95% CI, 0.33-0.97]; P = .04) (Figure and Table). The all-cause mortality rate was 8.1% in surgical group vs 8.6% in medical group (RR, 0.94 [95% CI, 0.46-1.94]; P = .93). No significant difference was shown in the per-protocol analysis for the primary outcome (18/149 [12.1%] vs 29/153 [19.0%]; RR, 0.64 [95% CI, 0.37-1.10]; P = .11) or for any secondary outcomes (Table). Sensitivity analyses are provided in the Table. When considering all possible outcome patterns in patients with missing data in a tipping-point analysis, 42.7% of the scenarios retained statistical significance. At the final follow-up, risk factor control in the surgical group vs the medical group was as follows: systolic pressure ≤130 mm Hg: 45/62 (72.6%) vs 53/75 (70.7%); P = .80; low-density lipoprotein cholesterol ≤1.81 mmol/L: 15/39 (38.5%) vs 22/60 (36.7%); P = .86; and hemoglobin A1c ≤7%: 23/25 (92.0%) vs 33/39 (84.6%); P = .38.

Discussion

This extended follow-up of CMOSS demonstrated that, although more composite outcome events (ie, stroke or death) happened within 30 days in the surgical group, there was a possible longer-term benefit in reducing the primary outcome among patients who received EC-IC bypass surgery plus medical treatment vs medical treatment alone. Compared with previous studies,2,3 the risk of recurrent stroke in the medical group remained high, with an increase from 12.3% at 2 years to 19.6% after a median of 7.6 years. Only 8 patients (5.3%) in the surgical group experienced the primary outcome beyond 2 years. The findings suggest that surgery, although conveying an up-front risk, may achieve long-term benefits if patients survive the perioperative danger. However, long-term follow-up of secondary outcomes, including any stroke, mortality, and the composite of any stroke or death, showed no significant differences.

Limitations of this study include the unblinded study design, nontrivial loss to follow-up, suboptimal stroke risk factor control, and violation of proportional hazards for the primary outcome. The findings should be interpreted cautiously given the post hoc long-term follow-up design, the neutral results of the preplanned analysis at 2 years, and the lack of statistical significance in other secondary outcomes.

The primary outcome was a composite of stroke or death within 30 days or ipsilateral ischemic stroke beyond 30 days after randomization. Data of patients lost to follow-up were treated as censored. The median (IQR) follow-up time was 7.57 (2.08-9.15) years for the surgical group and 7.69 (3.89-9.17) years for the medical group. The shading indicates 95% CIs of the primary outcome.

Figure.  Kaplan-Meier Plot of the Primary Outcome

Table.  Primary and Secondary Outcomes at Long-Term Follow-Up

이전 임상시험에서는, 혈류역학적으로 손상된 내경동맥(ICA) 또는 중대뇌동맥(MCA) 폐색 환자에서 두개외–두개내(EC-IC) 우회 수술이 내과적 치료보다 유의한 이점이 없다는 결과가 보고되어 왔다. CMOSS(Carotid and Middle Cerebral Artery Occlusion Surgery Study)는EC-IC 우회 수술 + 내과적 치료와 내과적 치료 단독을 비교한 다기관 무작위 임상시험이다. 초기 2년 추적 결과에서는 뚜렷한 차이가 없었으나, 이번 보고는 장기 추적 결과(최대 10년)를 포함하고 있다.

·         방법

▫ 초기 계획된 추적조사가 끝난 후, 각 연구센터의 기관윤리위원회 승인을 받아 최대 10년까지 장기 

추적이 가능하도록 연구 계획이 수정되었다.

▫ 추가 동의서가 필요했으며, 무작위 배정된 330명 중 324명이 장기 추적에 참여하였다

▫ 1차 결과(primary outcome): 무작위 배정 후 30일 이내 모든 뇌졸중 또는 사망, 혹은 30일 이 후 폐색된 동맥 측의 허혈성 뇌졸중 발생

▫ 통계분석: 로그-순위 검정, 카이제곱 검정, t검정 또는 윌콕슨 순위 검정, 카플란-마이어 분석 사용

·         결과

▫ 총 324명: 수술군 161명, 내과치료군 163명

▫ 중앙값 추적 기간: 7.6년 (IQR: 2.3~9.2년)

▫ 2년 추적 완료: 수술군 151명, 내과군 155명

▫ 장기 추적 손실: 수술군 24명, 내과군 22명

▫ 최종 완전한 데이터 보유: 255명(78.7%)

▫ 주요 결과 

-1차 결과 발생률:

수술군: 11.2% (18/161)

내과군: 19.6% (32/163)

상대위험도(RR): 0.57 (95% CI: 0.33–0.97), P = .04 → 유의미한 차이

-전체 사망률:

수술군: 8.1%

내과군: 8.6%

P = .93 → 유의미한 차이 없음

-Per-protocol 분석에서는 1차 결과 유의미한 차이 없음 (RR: 0.64, P = .11)

- 2차 결과 모두 유의미한 차이 없음

▫ 민감도 분석 

-실종 데이터에 대한 tipping-point 분석에서, 전체 시나리오 중 42.7%가 통계적 유의성 유지

▫ 위험인자 조절 (최종 시점 기준):

- 수축기 혈압 ≤130 mmHg: 수술군 72.6%, 내과군 70.7% (P = .80)

- LDL ≤1.81 mmol/L: 수술군 38.5%, 내과군 36.7% (P = .86)

- HbA1c ≤7%: 수술군 92.0%, 내과군 84.6% (P = .38)

·         논의 

 이번 CMOSS 장기 추적 분석에서는,

▫ 수술군에서 초기 30일 내 합병증 위험이 더 높았지만,

▫ 시간이 지나면서 수술군이 주요 복합 사건(뇌졸중 또는 사망) 발생률이 더 낮은 경향을 보여줌

▫ 내과 치료군은 2년간 재발률 12.3% → 7.6년간 19.6%로 증가,반면 수술군은 2년 이후 단 8명(5.3%)만이 주요 사건을 경험

이는 수술이 초기 위험을 감수하더라도 장기적으로 이점을 가질 수 있음을 시사함

그러나 사망률, 전체 뇌졸중 발생률, 2차 결과에서는 유의한 차이가 없음

·        연구의 한계

 ▫ 눈가림(블라인드) 처치 아님

▫ 추적 손실이 적지 않음

▫ 위험인자 조절이 최적이 아님

▫ 1차 결과의 비례위험 가정 위반

▫ 2년 분석에서는 중립적인 결과였고, 장기 추적은 사후 분석(post hoc)이므로 해석에 신중 필요