The New England Journal of Medicine June 11, 2009

A Randomized Trial of Therapies for Type 2 Diabetes and Coronary Artery Disease

BARI 2D Study Group (Frye RL, August P, Brooks MM, et al.)

Bottom Line

In patients with type 2 diabetes and stable coronary artery disease, a strategy of prompt revascularization did not significantly improve 5-year survival compared to intensive medical therapy alone, though a prespecified subgroup with severe disease benefited from CABG.

Key Findings

1. Overall Survival: At 5 years, the survival rate did not differ significantly between the prompt revascularization group (88.3%) and the intensive medical therapy group (87.8%) [P=0.97].

2. Major Cardiovascular Events: Freedom from the composite secondary endpoint of death, myocardial infarction (MI), or stroke at 5 years was similar between prompt revascularization (77.2%) and medical therapy (75.9%) [P=0.70].

3. CABG Stratum Benefit: In the prespecified subgroup of patients deemed appropriate for CABG due to more severe/multivessel disease (n=763), prompt revascularization significantly improved freedom from major cardiovascular events compared to medical therapy alone (77.6% vs. 69.5%, P=0.01). This benefit was primarily driven by a significant reduction in nonfatal MI (7.4% vs. 14.6%).

4. PCI Stratum: In the stratum of patients deemed appropriate for PCI, prompt revascularization did not improve survival or freedom from major cardiovascular events compared to medical therapy.

5. Glycemic Control Strategies: There was no significant difference in 5-year survival (88.2% vs. 87.9%, P=0.89) or freedom from major cardiovascular events (77.7% vs. 75.4%, P=0.13) between the insulin-sensitization and insulin-provision strategies.

Study Design

Design

RCT (2x2 Factorial)

Open-Label

Sample

2,368

Patients

Duration

5.3 yr

Median

Setting

Multicenter, multinational

Population Patients with type 2 diabetes and stable, angiographically documented coronary artery disease (≥50% stenosis with a positive stress test, or ≥70% stenosis with classic angina) who were candidates for elective PCI or CABG.

Intervention Prompt coronary revascularization (PCI or CABG chosen pre-randomization) plus intensive medical therapy; simultaneously randomized to either an insulin-sensitization or insulin-provision glycemic strategy.

Comparator Intensive medical therapy alone (with delayed revascularization only if clinically indicated); simultaneously randomized to either an insulin-sensitization or insulin-provision glycemic strategy.

Outcome All-cause mortality (5-year survival rate).

Study Limitations

• The trial was not designed to directly compare PCI against CABG; the revascularization method was chosen by the treating physician prior to randomization based on angiographic severity, preventing a head-to-head procedural comparison.

• The study excluded patients requiring immediate revascularization (e.g., left main disease, severe unstable angina), limiting generalizability to more acute presentations.

• The open-label design could introduce bias in patient management, particularly regarding the threshold for delayed revascularization in the medical therapy arm.

• Approximately 38% of patients randomized to the intensive medical therapy group ultimately crossed over to receive revascularization during the 5-year follow-up, which may have diluted the observable treatment effect in the intention-to-treat analysis.

• Advances in medical therapy and drug-eluting stents since the trial's enrollment period may affect the contemporary applicability of the specific event rates observed.

Clinical Significance

BARI 2D provided critical, nuanced evidence for managing stable coronary artery disease in diabetics. It reinforced the paradigm set by the COURAGE trial that, for most stable diabetic patients, intensive medical therapy is a highly effective and safe initial approach. It challenged the reflexive use of PCI to prevent mortality or MI in stable patients. Crucially, the trial identified a clear signal that diabetic patients with more extensive, complex multivessel disease experience fewer cardiovascular events with CABG. This finding subsequently guided international guidelines to heavily favor CABG over PCI for diabetics with multivessel disease, a strategy definitively validated later by the FREEDOM trial. Additionally, the endocrinologic arm demonstrated that insulin-providing regimens do not inherently worsen cardiovascular outcomes compared to insulin sensitizers, granting clinicians flexibility in tailoring glycemic control.

Historical Context

Prior to BARI 2D, the 2007 COURAGE trial had demonstrated that PCI offered no mortality or MI benefit over optimal medical therapy in stable CAD, but diabetic patients were underrepresented. Diabetics possess a distinct pathophysiology characterized by diffuse, rapidly progressive atherosclerosis. The original BARI trial (1996) had shown a survival benefit for CABG over balloon angioplasty in diabetic patients with multivessel disease. BARI 2D was specifically designed using a 2x2 factorial approach to definitively evaluate both revascularization strategies and the longstanding debate over whether insulin-providing therapies (implicated by the historical UGDP study as potentially harmful) increased cardiovascular risk compared to insulin-sensitizing agents.

Guided Discussion

High-yield insights from every perspective

Med Student

Medical Student

How does the pathophysiology of coronary artery disease in patients with type 2 diabetes differ from those without diabetes, and why might this influence the choice between CABG and PCI?

Key Response

Diabetes causes diffuse, multi-vessel, and accelerated atherosclerosis with smaller reference vessel diameters and greater plaque burden. This makes PCI less durable due to higher restenosis rates, explaining why the BARI 2D CABG stratum showed better outcomes for CABG compared to medical therapy or PCI in complex disease.

Resident

A 65-year-old patient with type 2 diabetes and stable angina is found to have multi-vessel coronary artery disease suitable for either PCI or CABG. Based on the BARI 2D trial, how should you counsel this patient regarding upfront revascularization versus an initial trial of intensive medical therapy?

Key Response

You should counsel that BARI 2D showed no overall survival benefit for prompt revascularization over intensive medical therapy (IMT) in stable CAD. However, if their disease severity specifically warrants CABG (e.g., complex multi-vessel disease), revascularization significantly reduces the rate of major cardiovascular events compared to IMT alone. Thus, a trial of IMT is very reasonable for less severe disease, while CABG should be strongly considered for complex multi-vessel disease.

Fellow

The BARI 2D trial stratified patients into PCI and CABG strata prior to randomization based on physician discretion. How does this non-randomized stratification impact the interpretation of the cardiovascular event reduction seen in the CABG arm versus the lack of benefit in the PCI arm?

Key Response

Because the choice of PCI vs. CABG stratum was non-randomized and based on angiographic severity, the baseline anatomical risk profiles of the two strata were fundamentally different. The CABG stratum had more severe, complex multi-vessel disease. Therefore, you cannot directly compare the efficacy of PCI versus CABG from this trial; you can only conclude that in patients whose severe disease makes them CABG candidates, prompt surgery beats medical therapy, whereas in patients whose less severe disease makes them PCI candidates, prompt PCI does not beat medical therapy.

Attending

When integrating the findings of BARI 2D with the more recent ISCHEMIA trial, how do you operationalize 'intensive medical therapy' in your clinic for diabetic patients with stable ischemic heart disease, and what are your clinical thresholds to declare medical therapy a failure?

Key Response

BARI 2D and ISCHEMIA both support starting with optimal medical therapy (OMT) for stable CAD. Today, OMT includes aggressive lipid lowering, antiplatelets, beta-blockers, ACEi/ARBs, and modern glycemic control with proven cardiovascular benefit (SGLT2 inhibitors and GLP-1 receptor agonists, which were not available during BARI 2D). Failure is defined clinically by refractory angina that unacceptably impairs quality of life despite maximal tolerated antianginal therapy, or progression to acute coronary syndrome, prompting crossover to revascularization.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD

The BARI 2D trial utilized a 2x2 factorial design to simultaneously evaluate revascularization strategies and glycemic control strategies (insulin sensitization vs. provision). What are the critical statistical assumptions required for a 2x2 factorial design to be valid, and how might an interaction between the two interventions affect the trial's power?

Key Response

A fundamental assumption of a 2x2 factorial design is the absence of interaction between the two interventions, meaning the effect of one intervention is consistent across the levels of the other. This allows independent estimation of marginal effects with full statistical power. If a synergistic or antagonistic interaction exists, the main effects analysis becomes confounded, and the study is typically heavily underpowered to detect this interaction, complicating the interpretation of the overall intervention efficacy.

Journal Editor

Nearly 20% of patients assigned to the intensive medical therapy arm in BARI 2D crossed over to undergo revascularization during the 5-year follow-up. As a reviewer, how does this crossover rate affect the interpretation of the intention-to-treat (ITT) analysis regarding the non-inferiority of medical therapy?

Key Response

High crossover rates in the medical therapy arm can bias the ITT results toward the null, potentially masking a true difference between the groups and making medical therapy appear falsely non-inferior. A rigorous statistical review would demand a per-protocol or as-treated sensitivity analysis to ensure that the observed lack of survival difference in the ITT analysis wasn't driven by medical therapy failures being 'rescued' by the revascularization intervention.

Guideline Committee

Based on the findings of BARI 2D, alongside subsequent trials like FREEDOM, how should current ACC/AHA guidelines grade the recommendation for CABG over PCI or medical therapy in diabetic patients with complex multi-vessel coronary artery disease?

Key Response

BARI 2D heavily informs current revascularization guidelines. The ACC/AHA guidelines give a Class 1 recommendation for CABG over PCI in diabetic patients with complex multi-vessel CAD (especially involving the LAD) to improve survival and reduce major adverse cardiovascular events. The committee must weigh the superiority of CABG in this specific BARI 2D subgroup against the general safety of initial medical therapy for less severe, stable disease, cementing CABG as the definitive standard of care for severe diabetic CAD.

Clinical Landscape

Noteworthy Related Trials

2007

COURAGE Trial

n = 2,287 · NEJM

Tested

PCI plus optimal medical therapy

Population

Patients with stable coronary artery disease

Comparator

Optimal medical therapy alone

Endpoint

Death from any cause and nonfatal myocardial infarction

Key result: As an initial management strategy in patients with stable CAD, PCI did not reduce the risk of death, MI, or other major cardiovascular events when added to optimal medical therapy.

2012

FREEDOM Trial

n = 1,900 · NEJM

Tested

Coronary artery bypass grafting (CABG)

Population

Patients with diabetes and multivessel coronary artery disease

Comparator

Percutaneous coronary intervention (PCI) with drug-eluting stents

Endpoint

Composite of death from any cause, nonfatal myocardial infarction, or nonfatal stroke

Key result: CABG significantly reduced rates of death and myocardial infarction compared to PCI, though with a higher rate of stroke.

2020

ISCHEMIA Trial

n = 5,179 · NEJM

Tested

Routine invasive strategy (angiography plus revascularization)

Population

Patients with stable coronary disease and moderate or severe ischemia

Comparator

Conservative strategy (medical therapy alone)

Endpoint

Composite of CV death, MI, resuscitated cardiac arrest, or hospitalization for unstable angina or heart failure

Key result: An initial invasive strategy did not demonstrate a reduced risk of ischemic cardiovascular events or death compared to a conservative strategy.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis