Review of the Coronary Artery Bypass Graft Surgery Trialists Collaboration

A patient-level meta-analysis of coronary artery bypass surgery in stable angina

Lancet 1994;344(8922):563-570

Background: Several randomized trials have assessed the benefits of coronary artery bypass graft surgery (CABG) in patients with chronic stable angina. We reviewed three major trials in this area: The Veterans Administration Cooperative Study, the European Coronary Surgery Study, and the CASS study. These trials yielded divergent results, likely due to differences in patients included and surgical conduits used.

The aim of this study was to conduct a patient-level meta-analysis of these three major trials, plus four additional smaller trials, to compare the long-term effects, at 10 years, of coronary artery bypass surgery versus medical therapy.

Note to readers: We are including this meta-analysis among our reviews since it’s frequently referenced when discussing the benefits of coronary artery bypass surgery.

Patients: Patients were included from 7 trials of coronary artery bypass surgery in chronic stable angina. Patients entered the trials between 1972 and 1984. Left ventricular systolic function was considered abnormal if the ejection fraction was < 50%. Coronary artery was considered to be diseased if it had a stenosis > 50%. Angiographic data were available from all trials except one of the four smaller trials.

Baseline characteristics: The study included a total of 2,649 patients; 1,324 patients were assigned to receive surgery and 1,325 were assigned to receive medical therapy. The three major trials contributed 2,233 (84.3%) patients while the four smaller trials combined contributed 416 (15.7%) patients. The average age of patients was 51 years with 97% being men. The mean ejection fraction was 59%. About 60% had prior myocardial infarction, 26% had hypertension, 4% had heart failure, 10% had diabetes, and 84% had smoking history.

Approximately 10% had single vessel disease, 32% had two-vessel disease, and 51% had three-vessel disease. Left main disease was present in about 7% of the patients, and about 59% had proximal left anterior descending artery disease.

Procedures: Individual patients’ data were collected on a standardized form. Patients in the trials were randomly assigned to coronary artery bypass surgery or medical therapy.

Endpoints: The primary end point was all-cause death. The statistical analysis was performed based on the intention-to-treat principle, irrespective of crossover. Authors also sought to assess the interaction of treatments based on the extent of coronary artery disease, left ventricular dysfunction and some comorbid conditions.

Results: The authors reported mortality at 5, 7 and 10 years. In this review, we will focus on the 10 years results as this was the intention of the study and most patients had 10-year follow-up data available. Of those assigned to surgery, 93.7% underwent the procedure, while 37.4% of patients initially assigned to medical therapy ended up having surgery. Among the patients assigned to surgery, the 30-day operative mortality was 3.2%. The mean number of grafts used was 2.4 and 9.9% of the patients received an internal mammary artery graft.

Antiplatelets were used by 18.8% of the patients in the medical arm and 25.5% in the surgical arm.

CABG significantly reduced mortality compared to medical therapy alone (26.4% vs 30.5%, OR: 0.83, 95% CI: 0.70 – 0.98; p= 0.03). CABG reduced mortality in patients with left main disease (OR: 0.67; p= 0.24 “this is insignificant due to small sample size”) and in patients with three vessel disease (OR: 0.76; p=0.02). CABG also reduced mortality in patients with one or two vessel disease if they had left anterior descending artery disease (OR and p value for 10-years were not provided). Absolute values were not provided for any of these subgroups at 10-years.

Authors reported no significant interactions based on history of hypertension, smoking or prior myocardial infarction (no event rate or results of interaction testing were provided for these groups). At 5-years, patients with left ventricular dysfunction derived similar benefit with surgery compared to patients with normal left ventricular function (p for interaction= 0.90).

Conclusion: In patients with stable angina pectoris and > 50% stenosis in at least one of the major coronary arteries, coronary artery bypass surgery compared to medical therapy alone reduced death at 10-years with a number need to treat of approximately 24 patients. The benefit was seen in patients with left main disease, three vessel disease and patients with left anterior descending artery disease. No benefit was seen in patients with one or two vessel disease if they did not have left anterior descending artery disease.

Of note, unlike most current meta-analyses that rely on trial-level data, this type of meta-analysis provides significantly higher level of evidence as it uses patient-level data. This method allows for the categorization of patients based on uniform definitions, the ability to reperform the analysis and helps address some limitations in previous trials. For example, the current study used an intention-to-treat analysis, while the Veterans Administration Cooperative study used a "modified intention-to-treat analysis."

Despite these strengths, some limitations from the original trials carry over to this study such as the small number of patients with left main disease (about 7% of study population) and the infrequent use of internal mammary artery (about 10%). Furthermore, the advancement in both medical therapies as well as surgical techniques since the publication of this study, limit its external validity in the current era.