By Jeffrey Zimmet, MD, PhD

Associate Professor of Medicine, University of California, San Francisco; Director, Cardiac Catheterization Laboratory, San Francisco VA Medical Center

Dr. Zimmet reports no financial relationships relevant to this field of study.

SYNOPSIS: The NOBLE and EXCEL trials randomized patients presenting with left main disease to treatment with percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG). NOBLE reported an advantage to CABG at five years, while EXCEL demonstrated non-inferiority of PCI compared with CABG at three years.

SOURCES: Stone GW, Sabik JF, Serruys PW, et al. Everolimus-eluting stents or bypass surgery for left main coronary artery disease. N Engl J Med 2016 Oct 31. [Epub ahead of print].

Mäkikallio T, Holm NR, Lindsay M, et al. Percutaneous coronary angioplasty versus coronary artery bypass grafting in treatment of unprotected left main stenosis (NOBLE): A prospective, randomised, open-label, non-inferiority trial. Lancet 2016 Oct 31. pii: S0140-6736(16)32052-9. doi: 10.1016/S0140-6736(16)32052-9. [Epub ahead of print].

Coronary artery bypass graft (CABG) has been the first-line treatment for significant disease involving the left main coronary artery for a long time. However, as stent technology and techniques have advanced, percutaneous coronary intervention (PCI) has become a viable option for a subset of these patients. Two large randomized trials comparing PCI to CABG for left main disease were presented recently at the 2016 TCT conference, and have subsequently been published. The media have focused on the supposedly contradictory conclusions of these two trials. Wherein lies the truth?

The NOBLE trial, conducted at 36 centers in Europe between late 2008 and early 2015, enrolled 1,201 patients presenting with left main coronary disease and randomized them to PCI or CABG. The primary endpoint was the Kaplan-Meier five-year estimate of a composite of major adverse cardiac and cerebrovascular events (MACCE; death from any cause, non-procedural myocardial infarction, repeat revascularization, or stroke), based on a median of three years of follow-up. Eleven percent of the PCI group, enrolled in the early phase of the trial, were treated with older first-generation drug-eluting stents. Thereafter, the recommended study stent became a biolimus-eluting stent that is available in Europe but not in the United States. The five-year estimate of MACCE occurred in 29% of patients treated with PCI vs. 19% of those who underwent CABG, a difference that exceeded the limit for noninferiority (P = 0.007 for superiority). Although mortality rates were not different between the groups, the rate of myocardial infarction (MI) was significantly higher in the PCI group (6.9% vs. 1.9%; hazard ratio [HR], 2.88; 95% confidence interval [CI], 1.40-5.90), as was any revascularization (16.2% vs. 10.4%; HR, 1.50; 95% CI, 1.04-2.17). Interestingly, stroke rates were numerically higher in the PCI group as compared to CABG, although this did not quite reach statistical significance at five years. Importantly, periprocedural MIs occurring in the first 30 days were excluded from the analysis. The NOBLE investigators concluded that CABG “might be better than PCI” for treatment of left main coronary disease.

The EXCEL trial began with a similar premise, but with several major differences. EXCEL, starting nearly two years later than NOBLE in late 2010, randomized 1,905 left main patients at 126 centers in 17 countries to PCI or CABG. Because the SYNTAX trial had suggested poorer performance of PCI as compared to CABG in patients with higher disease complexity as assessed by the SYNTAX score, the trial aimed to enroll only patients with low ( 22) or intermediate (23-32) SYNTAX scores. Patients presented with an average SYNTAX score of 20.5, which was slightly lower than the population in NOBLE (average score: 22). All PCI patients received the second-generation Xience everolimus-eluting stent.

The primary endpoint, a composite of all-cause death, MI, and stroke at three years, occurred in 15.4% of the patients in the PCI group and in 14.7% of the patients in the CABG group (95% CI, 0.79-1.26; P = 0.02 for noninferiority). Although ischemia-driven revascularization during follow-up was more frequent after PCI than after CABG (12.6% vs. 7.5% of patients, P < 0.001), the secondary composite endpoint of death, stroke, MI, or ischemia-driven revascularization, which more closely mirrors the primary endpoint from NOBLE, showed a trend favoring CABG but was not statistically significant (23.1% vs. 19.1%; P = 0.10). Unlike NOBLE, the EXCEL investigators included periprocedural MI in the study data and analyzed 30-day outcomes as a separate secondary endpoint. The rate of death, stroke, or MI at 30 days was significantly higher among the CABG cohort (4.9% PCI vs. 7.9% CABG; HR, 0.61; 95% CI, 0.42-0.88), driven primarily by a significantly increased risk of MI (3.9% vs. 6.2%; HR, 0.63; 95% CI, 0.42-0.95). Notably, the designation of MI included a programmed measurement of CK-MB post procedure, with values above 10 times the upper reference limit meeting the definition for MI, even in the absence of other clinical features.

The EXCEL investigators concluded that in patients with left main disease and low or moderate overall complexity of disease as defined by the SYNTAX score, PCI with everolimus-eluting stents was non-inferior to CABG at three years of follow-up. In the accompanying editorial, Eugene Braunwald summed up the study by saying that “the majority of patients with left main coronary disease ... can now be managed equally well by means of two strategies of revascularization if carried out by expert, experienced teams.”


Before getting into the obvious question — how two large randomized trials of the same question could come to seemingly opposite conclusions — a few points are worth making. First, it is important to step back and realize that not all patients with left main disease are equally treatable by both revascularization techniques. In the initial enrollment period of EXCEL, for example, out of 1,747 patients, only 1,078 (61.7%) were determined to be eligible for PCI, and 1,395 (79.9%) were eligible for CABG. Evaluation by experienced teams of interventional cardiologists and surgeons is necessary in every case. Critics have highlighted several issues with the NOBLE trial. The study was relatively underpowered, and the primary endpoint changed at least twice from the original design because of a low event rate. It used a significant number of older-generation stents, and had a lower rate of intravascular ultrasound use (compared with EXCEL), which both studies encouraged for all left main interventions. Nevertheless, the primary finding of a CABG advantage driven by lower rates of subsequent revascularization and non-procedure-related MI is plausible and matches the existing body of evidence.

For EXCEL, much of the discussion and uncertainty centers on the use of the 30-day outcomes, including periprocedural enzyme-defined MIs, and on the relatively short duration of follow-up. On the former issue, one can argue that the periprocedural MI rate drove the primary endpoint to non-inferiority. Some have argued that exclusion of these early events would push the primary outcome in favor of CABG. Indeed, examination of the event curves suggests a catch-up in the PCI group after the initial 30-day outcomes, and a potential separating of curves in the CABG direction with more time. Longer-term follow-up of the study population out to five years and longer will be important to put some of these concerns to rest, but for now, true long-term comparative data are lacking.

These two trials represent the best comparative left main data we are likely to have for the foreseeable future. Both trials demonstrate that left main PCI in patients with low or intermediate lesion complexity as defined by the SYNTAX score, when performed in the context of a cooperative heart team at experienced centers, can be performed with a high degree of safety and with good outcomes out to at least three years. There is a clear advantage of PCI in the first 30 days, which appears to be balanced by more clinical events over time compared with CABG. For now, patients who are good candidates for surgery are likely to continue to be referred for CABG evaluation. However, patient preference almost certainly will play an increasing role in determining the mode of revascularization in cases in which either approach is technically feasible, and guidelines are likely to reflect this trend.