EXCEL at 5 Years: PCI, CABG 'Relatively Comparable' for Left Main Coronary Disease

September 30, 2019

SAN FRANCISCO — Five-year outcomes from a major randomized trial provide at least a partial answer to whether coronary stenting for left main (LM) disease may be better than coronary artery bypass grafting (CABG), or vice versa, for preventing serious clinical outcomes over the long haul.

Surgery and percutaneous coronary intervention (PCI) were on par with each other for death from any cause, myocardial infarction (MI), or stroke in the EXCEL trial's final 5-year analysis.

The finding is similar to its prospectively defined 3-year primary outcome for the same composite endpoint. As previously reported, EXCEL showed PCI with the everolimus-eluting Xience stent (Abbott Vascular) to be noninferior to CABG for treating LM coronary lesions of low- to mid-range complexity.

The less invasive procedure showed an early advantage, but CABG caught up later for composite endpoint rates of 15.4% and 14.7%, respectively.

But the extra 2 years of follow-up cast those earlier findings in a different light: a short-term significant advantage for PCI over CABG had turned into a tie at about 3 years, but the trajectories continued such that CABG fared better in later years.

"Early on there was a clear benefit for PCI, with fewer periprocedural events, mostly due to fewer large MIs and somewhat less strokes," said Gregg W. Stone, MD, Icahn School of Medicine at Mount Sinai, New York City, when outlining the EXCEL 5-year outcomes for journalists here at Transcatheter Cardiovascular Therapeutics (TCT) 2019.

The risk reduction with PCI reached 39% at 30 days (P = .008). But events were fewer after about 1 year for those who had undergone CABG, for a 61% risk increase with PCI starting at 1 year out to 5 years.

"The early benefits of PCI of reduced periprocedural risk were attenuated by the greater number of events occurring at follow-up than with bypass surgery, such that at 5 years the accumulated mean time free from adverse events was similar with both treatments," he said.

"PCI may thus be considered an acceptable revascularization modality for selected patients with left main disease, a decision which should be made after heart team discussion taking into account each patient’s individual risk factors and preferences," Stone concluded for reporters, similarly to his wrap-up after formally presenting the results here at TCT 2019.

Stone is also lead author on the same-day publication of the EXCEL 5-year outcomes in the New England Journal of Medicine.

"To me the signal was the same no matter how you looked at it, that PCI was good in the short term, and that CABG seemed to be better in the longer term," interventionalist Dharam J. Kumbhani, MD, University of Texas Southwestern Medical Center, Dallas, told theheart.org | Medscape Cardiology.

As to the message for practice, "If you put this in context with other trials that have been done, NOBLE and others, I would think the field would probably favor the longer-term benefit with CABG," said Kumbhani, who wasn't part of EXCEL.

But it would depend on the patient's condition and preferences, he said, and one would have to consider the variable benefits in EXCEL from both PCI and CABG for a range of individual types of events.

For example, Kumbhani said, the two treatments were associated with similar rates of MI, but periprocedural MIs were significantly less likely with PCI; all other types of MI were less likely with CABG. "It's hard to know, but downstream these patients may have a higher MI risk."

Also, PCI protected better against cerebrovascular events, but was followed by more ischemia-driven coronary revascularization over 5 years.

But these are only signals, he observed, "so I think it's a very challenging discussion."

Stone also noted that all-cause mortality as a solo endpoint was significantly increased over 5 years, by 38%, in the PCI arm. But, he added at the media briefing, "Most of the difference was in noncardiovascular mortality. This was because of late infections and late sepsis that were unrelated to any of the procedures and likely due to a play of chance."

Table 1. Odds Ratio (OR) for 5-Year Primary and Secondary Outcomes in EXCEL, PCI vs CABG
End points OR (95% CI)
Death, Stroke, or MI 1.19 (0.95 - 1.50)
Death From Any Cause 1.38 (1.03 - 1.85)
Cerebrovascular Events 0.61 (0.38 - 0.99)
All MI 1.14 (0.84 - 1.55)
Periprocedural MI 0.63 (0.41 - 0.96)
Nonperiprocedural MI 1.96 (1.25 - 3.06)
Ischemia-driven Coronary Revascularization 1.84 (1.39 - 2.44)
Abbreviations: CABG, coronary artery bypass grafting; CI, confidence ratio; OR, odds ratio; MI, myocardial infarction; PCI, percutaneous coronary intervention.

The analysis wasn't powered for such individual secondary endpoints, and the noncardiovascular death increase in EXCEL's PCI arm was likely just by chance, Stone proposed when presenting the results formally.

"But I think most importantly, these are pretty small differences — we're talking 2% or 5%," he said, which are minimal rates per year over 5 years.

"So I think we need to have perspective," he said. "They're relatively comparable therapies that lead to overall comparable large major clinical outcomes."

The 5-year EXCEL outcome that most matters is the primary composite of death, stroke, or MI for which the trial was powered. It was increased 19% in the PCI group, said Stuart J. Pocock, PhD, London School of Hygiene and Tropical Medicine, England, who spoke as a panelist following Stone's formal presentation of the trial.

"It's not significant, but you cannot rule out the possibility of a 20% excess," said Pocock, a statistician and expert on clinical trial design.

"The real challenge is interpreting the mortality results," he said, given the excess of noncardiovascular deaths in the PCI group "in a way that was pretty illogical."

Therefore, Pocock said, "I think the mortality finding is a concern, but to me it's plausible that it is a chance finding" when also considering other trial evidence.

"If you were an operator who says every patient with left main disease should just have PCI, I think this would probably temper the enthusiasm," Kumbhani said about EXCEL in an interview.

On the other hand, for a more informed decision in collaboration with a patient, according to this study, "at 5 years, your risk profile would probably be different, but if you look at the totality of endpoints that you care about, whether you live or whether you die, or whether you have a stroke or heart attack, you will do just as well with PCI as with CABG," he said.

EXCEL had randomly assigned 1905 patients with unprotected LM coronary disease, SYNTAX disease complexity scores 32 or lower, and no prior LM revascularization to undergo PCI with Xience or CABG, 948 and 957 respectively, at 126 sites in 17 countries.

At both baseline and at 5 years (93% PCI follow-up and 90% CABG follow-up), those who had surgery were significantly more likely to be on chronic oral anticoagulation, beta blocker, diuretic, and antiarrhythmic medications, and less likely to be on P2Y12-inhibitor antiplatelets or either ACE inhibitors or angiotensin receptor blockers.

Table 2. Hazard Ratio (HR) or Odds Ratio (OR) for Outcomes in EXCEL, PCI vs CABG
End points HR/OR (95% CI) P value
3 Years: Death, Stroke, or MI 1.00 (0.79 - 1.26) .98
5 Years: Death, Stroke, or MI 1.19 (0.95 - 1.50) .13
5 Years: Death, Stroke, MI, or Ischemia-driven Revascularization 1.39 (1.13 - 1.71) .002

Stone also presented a "piecewise hazards" analysis that illustrated the pattern of differences in the 3-year composite primary endpoint within three discrete segments of the long-term follow-up.

In the perioperative period up to day 30, the rate was significantly lower with PCI. From day 30 to 1 year, the period of highest risk for stent restenosis, Stone pointed out, there was no significant difference between the two arms. And the risk was significantly increased for PCI compared with CABG from 1 year through 5 years.

Table 3. Piecewise Hazard Ratios (HR) for Death From Any Cause, Stroke, or MI in EXCEL, PCI vs CABG
End Points HR (95% CI) P valuea
Up to Day 30 0.61 (0.42 - 0.88) .008
Day 30 to 1 Year 1.07 (0.68 - 1.70) .76
1 Year to 5 Years 1.61 (1.23 - 2.12) < .001
a P value for treatment-time interaction < .001

In EXCEL, PCI was best in the short term and less effective than CABG in the later years of follow-up, as is often the case with less-invasive procedures compared with their more-invasive counterparts, observed surgeon Kenneth Ouriel, MD, MBA, president of Syntactx, New York City, who was not involved in the study.

But it remains that PCI and CABG fared similarly for the primary endpoint over the 5 years, and "in my experience, if you can show the things are the same, patients are going to go for the less invasive procedure almost 100% of the time," he told theheart.org | Medscape Cardiology.

On the other hand, if the patient is younger, say 65 rather than 80, "then you have to think that the late part of that curve matters more."

EXCEL was sponsored by Abbott Vascular. Stone, Kumbhani, and Pocock have disclosed to relevant financial relationships. Ouriel disclosed equity, stock, or options interest in and a salary from Syntactx.

Transcatheter Cardiovascular Therapeutics 2019: Late Breaking Trials 3. Presented September 28, 2019.

New Engl J Med. Published online September 28, 2019. Abstract

Follow Steve Stiles on Twitter: @SteveStiles2. For more from theheart.org | Medscape Cardiology, follow us on Twitter and Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.