New England Journal of Medicine DECEMBER 08, 2016

Everolimus-Eluting Stents or Bypass Surgery for Left Main Coronary Artery Disease

Gregg W. Stone, Joseph F. Sabik, Patrick W. Serruys, et al.

Bottom Line

In patients with left main coronary artery disease of low or intermediate anatomical complexity, PCI with everolimus-eluting stents was noninferior to CABG for the composite endpoint of death, stroke, or myocardial infarction at 3 years.

Key Findings

1. At3years, theprimarycompositeendpointofdeath, stroke, ormyocardialinfarctionoccurredin15.4%ofpatientsinthePCIgroupand14.7%intheCABGgroup(HazardRatio1.00;95%CI0.79–1.26;P=0.02fornoninferiority)[1.2].
2. At 30 days, the composite rate of death, stroke, or MI was significantly lower in the PCI group (4.9%) compared to the CABG group (7.9%) (P=0.008 for superiority), largely due to a reduction in periprocedural MI and early adverse events.
3. The secondary composite endpoint of death, stroke, MI, or ischemia-driven revascularization at 3 years occurred in 23.1% of the PCI group versus 19.1% of the CABG group.

Study Design

Design
RCT
Open-Label
Sample
1,905
Patients
Duration
3 yr
Median
Setting
Multicenter, international
Population Patients with left main coronary artery disease and low or intermediate anatomical complexity (site-assessed SYNTAX score ≤ 32).
Intervention Percutaneous coronary intervention (PCI) with fluoropolymer-based cobalt-chromium everolimus-eluting stents.
Comparator Coronary-artery bypass grafting (CABG).
Outcome A composite of death from any cause, stroke, or myocardial infarction at 3 years.

Study Limitations

Thetrialwasopen-label, whichcanintroducebiasinthediagnosisofperiproceduraleventsandthedecisionthresholdforischemia-drivenrevascularization.
Thetrialexclusivelyenrolledpatientswithloworintermediateanatomicalcomplexity(SYNTAXscore≤32), limitinggeneralizabilitytopatientswithhighlycomplexmultivesseldiseasewhoremainbettersuitedforCABG[1.2].
The 3-year follow-up for the primary outcome was relatively short; surgical benefits (such as graft patency and survival advantage) tend to manifest later in longitudinal follow-up.
The trial's non-standard definition of periprocedural MI was heavily debated by surgical societies as potentially favoring PCI over CABG.

Clinical Significance

The EXCEL 3-year results provided pivotal evidence supporting the use of second-generation everolimus-eluting stents as an acceptable, noninferior alternative to CABG for left main coronary artery disease in patients without highly complex coronary anatomy, directly influencing international revascularization guidelines. However, significant divergence in outcomes over time necessitates careful, individualized Heart Team discussions for optimal patient selection.

Historical Context

Historically, CABG was the undisputed gold standard for unprotected left main coronary artery (LMCA) disease. First-generation drug-eluting stent trials (like SYNTAX) showed PCI could be a viable alternative for less complex disease, albeit with higher repeat revascularization rates. The EXCEL trial evaluated contemporary, second-generation everolimus-eluting stents specifically in LMCAD patients with low-to-intermediate SYNTAX scores. While the 2016 3-year EXCEL results established noninferiority for PCI, the 2019 5-year follow-up generated unprecedented controversy by revealing a higher all-cause mortality rate for PCI (13.0%) compared to CABG (9.9%). This late mortality signal, combined with methodological debates over the trial's definition of periprocedural MI, prompted the European Association for Cardio-Thoracic Surgery (EACTS) to formally withdraw support for joint clinical guidelines that had equally recommended PCI for lower-complexity LMCAD.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

Why is significant stenosis in the left main coronary artery historically considered an absolute indication for coronary artery bypass grafting rather than percutaneous coronary intervention?

Key Response

The left main coronary artery supplies approximately 75 percent of the left ventricular myocardial mass. Abrupt closure during PCI or due to stent thrombosis can lead to catastrophic hemodynamic collapse and sudden cardiac death, which historically made CABG the safer and more durable option for revascularization.

Resident
Resident

In applying the findings of the EXCEL trial to a patient with left main disease, how does the SYNTAX score influence your decision between PCI and CABG?

Key Response

The EXCEL trial specifically enrolled patients with low or intermediate anatomical complexity, defined as a SYNTAX score of 32 or less. Patients with high complexity (SYNTAX score greater than 32) were excluded because CABG remains the absolute standard of care for highly complex, diffuse multi-vessel disease due to significantly better long-term survival and freedom from repeat revascularization.

Fellow
Fellow

How does the definition of peri-procedural myocardial infarction used in the EXCEL trial compare to the Universal Definition of MI, and why is this distinction critical when comparing PCI to CABG?

Key Response

EXCEL used the SCAI definition for peri-procedural MI, which requires larger biomarker elevations (e.g., CK-MB greater than 10x upper limit of normal) compared to the Universal Definition. Because CABG inherently causes more minor biomarker leaks than PCI due to myocardial manipulation and cardioplegia, using the Universal Definition would heavily penalize CABG. The choice of definition significantly influenced the composite primary endpoint and sparked widespread debate regarding the true clinical equivalence of the procedures.

Attending
Attending

Given the non-inferiority of PCI at 3 years in EXCEL but differing results in the NOBLE trial and subsequent 5-year mortality signals, how should the Heart Team approach shared decision-making with a 60-year-old diabetic patient presenting with isolated left main disease?

Key Response

The Heart Team must synthesize conflicting data and emphasize temporal risk trade-offs. While EXCEL showed non-inferiority at 3 years, CABG typically offers superior long-term durability, especially in diabetics. Shared decision-making must weigh the early risks of CABG (higher upfront bleeding and surgical recovery) against the late risks of PCI (higher rates of repeat revascularization and a potential mortality disadvantage at 5 years), tailoring the choice to the patient's surgical risk and personal values.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

The EXCEL trial utilized a non-inferiority design for its primary composite endpoint of death, stroke, or MI. How does the inclusion of differently weighted events in a composite endpoint threaten the validity of a non-inferiority margin?

Key Response

A composite endpoint treats all components as statistically equal. If a non-inferiority margin is met largely due to a reduction in a less severe, higher-frequency component (like biomarker-defined peri-procedural MI) while a more severe, lower-frequency component (like all-cause mortality) trends in the opposite direction, the statistical conclusion of non-inferiority may not translate to true clinical equivalence. This requires a rigorous analysis using weighted endpoints or competing risk models.

Journal Editor
Journal Editor

As a peer reviewer analyzing the EXCEL trial manuscript, what concerns would you raise regarding the decision not to publish the Universal Definition of MI data alongside the SCAI definition data in the primary manuscript?

Key Response

A critical reviewer would flag selective outcome reporting as a major threat to validity. Given that the trial's protocol pre-specified the collection of data using the Universal Definition of MI, failing to report it when it potentially favors the control arm (CABG) introduces significant bias. Furthermore, the reliance on an industry-sponsored, modified endpoint for a non-inferiority claim demands strict editorial scrutiny regarding transparency and methodological rigor.

Guideline Committee
Guideline Committee

How do the 3-year results of the EXCEL trial, when contrasted with its 5-year follow-up data and the independent NOBLE trial, impact the ACC/AHA and ESC/EACTS guideline recommendations (Class of Recommendation and Level of Evidence) for left main revascularization?

Key Response

Current guidelines generally give CABG a Class I recommendation for left main disease. Based on the 3-year EXCEL data, PCI received a Class IIa recommendation for low-to-intermediate SYNTAX scores. However, discordant results from NOBLE and the late mortality signal in EXCEL's 5-year data led the European Association for Cardio-Thoracic Surgery (EACTS) to officially withdraw support for the left main chapter of the ESC guidelines. Guideline committees must now debate whether to downgrade the PCI recommendation to Class IIb or Class III for intermediate complexity based on long-term safety signals.

Clinical Landscape

Noteworthy Related Trials

2009

SYNTAX Trial

n = 1,800 · NEJM

Tested

PCI with paclitaxel-eluting stents

Population

Patients with left main or three-vessel coronary artery disease

Comparator

CABG

Endpoint

Major adverse cardiac or cerebrovascular events (MACCE) at 12 months

Key result: CABG resulted in lower rates of MACCE at 1 year compared to PCI, primarily driven by a higher rate of repeat revascularization in the PCI group.
2011

PRECOMBAT Trial

n = 600 · NEJM

Tested

PCI with sirolimus-eluting stents

Population

Patients with unprotected left main coronary artery stenosis

Comparator

CABG

Endpoint

MACCE (death, MI, stroke, or ischemia-driven target-vessel revascularization) at 1 year

Key result: PCI was found to be non-inferior to CABG for the primary composite endpoint at 1 year, though ischemia-driven target-vessel revascularization was more frequent with PCI.
2016

NOBLE Trial

n = 1,201 · Lancet

Tested

PCI with biolimus-eluting stents

Population

Patients with unprotected left main coronary artery disease

Comparator

CABG

Endpoint

MACCE (death, non-procedural MI, repeat revascularization, stroke) at a median of 3 years

Key result: CABG was significantly better than PCI for the primary composite endpoint, driven by lower rates of non-procedural MI and repeat revascularization.

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