No Mortality Difference Between PCI and CABG for Left Main CAD: Meta-analysis

An eagerly awaited review is spurring calls to stop bickering and agree on the “facts” about left main revascularization.

No Mortality Difference Between PCI and CABG for Left Main CAD: Meta-analysis

The risk of death was no different following revascularization with CABG surgery or PCI for the treatment of left main coronary artery disease of mostly low-to-intermediate complexity, according to the results of an eagerly anticipated meta-analysis.

Presented today at the virtual American Heart Association 2021 Scientific Sessions and published simultaneously in the Lancet, the 5-year Kaplan-Meier estimate of all-cause mortality was 11.2% among those treated with CABG surgery and 10.2% for those treated with PCI, a difference that wasn’t statistically significant. In a Bayesian analysis, researchers reported there was a greater likelihood of more deaths from any cause at 5 years with PCI, although that excess risk was “more likely than not” less than 1% over 5 years, or less than 0.2% year. For cardiovascular mortality, the excess risk with PCI was likely less than 0.1% per year compared with surgery.   

Lead investigator Marc Sabatine, MD (Brigham and Women’s Hospital, Boston, MA), highlighted the importance of the new analysis given the dispute between the surgical and interventional communities in the treatment of left main CAD.

“The hope is that they can now sit down and all agree that we have the same facts in front of us,” Sabatine told TCTMD. “Our goal with putting all the trials together was to look at individual outcomes, starting with all-cause mortality, and looking at cardiovascular mortality, MI, stroke, and revascularization, and to look at events early and late, and to do a Bayesian analysis. At least for mortality, it wouldn’t be a simple dichotomous answer—‘Yes, it’s significantly different’ or ‘No, it’s not.’ We wanted to give people a sense of the probability and likelihood of any magnitude of difference that might exist between the two treatment arms.”   

The debate over the optimal revascularization approach for left main CAD has been simmering for years, with the mortality controversy first gaining steam with the EXCEL trial 5-year outcomes showing that PCI was associated with a higher risk of all-cause mortality compared with PCI. An ugly spat arose when David Taggart, MD, PhD (University of Oxford, England), chairman of the EXCEL surgical committee during the design and recruitment phase of the trial, split with the other trialists, in part over the mortality results and missing data on periprocedural infarctions. Shortly after the 5-year results of EXCEL were published, the European Association for Cardio-Thoracic Surgery (EACTS) withdrew their support from the European Society of Cardiology (ESC) revascularization guidelines for left main CAD and called for an independent review.

The hope is that they can now sit down and all agree that we have the same facts in front of us. Marc Sabatine

The meta-analysis, which includes senior author Eugene Braunwald, MD (Brigham and Women’s Hospital), was facilitated by the ESC to review the evidence on left main revascularization and help inform the revascularization guidelines, said Sabatine.

Four Major Left Main Trials   

In the European guidelines PCI is a class I indication for patients with a low SYNTAX score and a class IIa recommendation for those with an intermediate SYNTAX score. PCI is not recommended for patients with highly complex anatomy (SYNTAX score ≥ 33). In contrast, CABG surgery has a class I indication in all patients with left main CAD regardless of anatomical complexity. In the US guidelines, PCI is class IIa indication for left main CAD in patients with a low SYNTAX score and a class IIb indication for those with an intermediate SYNTAX score. Here again, CABG is indicated in left main patients regardless of anatomical complexity.      

The new meta-analysis included the four main revascularization trials of patients with left main CAD: EXCEL, NOBLE, SYNTAX (left main cohort only), and PRECOMBAT, all of which had at least 5 years of clinical follow-up. In total, the meta-analysis included 4,394 patients with a median SYNTAX score of 25.0 who were randomly assigned to PCI or CABG surgery. Between 21% and 22% of patients revascularized had a SYNTAX score ≥ 33.

At 5 years, the absolute difference in all-cause mortality between two revascularization strategies was 0.9% favoring CABG surgery, a difference that was not statistically significant. In a landmark analysis, all-cause mortality was lower with PCI in the first year after randomization, but trended higher between 1 and 5 years. During the latter time point, all-cause mortality was 8.7% for the PCI-treated patients and 7.2% for those treated with surgery (HR 1.22; 95% CI 0.98-1.52). At 10 years, an analysis based on PRECOMBAT and SYNTAX alone (the only trials with enough years of follow-up), the mortality rate was not significantly different between the PCI- and CABG-treated patients.   

Clinical Outcomes at 5 Years




P Value

All-Cause Mortality

    CV Death

    Non-CV Death










Spontaneous MI



< 0.001

Procedural MI (Protocol Definition)




Procedural MI (Universal Definition)




Any MI








Coronary Revascularization



< 0.001

Among those treated with PCI, the risk of spontaneous MI was significantly higher compared with surgery, as was the need for repeat revascularization. Depending on the definition used, the rate of procedural MI varied, favoring PCI when the study’s protocol definitions were used but not when the Universal Definition captured events. Stroke rates did not significantly differ at 5 years, although the risk was significantly lower with PCI in the first year after randomization. 

Sabatine noted that the field has been mired in debate over the risks of MI with each procedure, and how best to define them. In EXCEL, for example, the researchers were criticized for their use of a modified Society for Cardiovascular Angiography and Interventions (SCAI) definition for periprocedural MI and for initially not reporting MI events captured using the Universal Definition. For that reason, Sabatine said it was important to tease out spontaneous MIs from procedural infarctions using different definitions. 

While Sabatine, speaking with TCTMD, didn’t want to wade into the guideline recommendations for left main CAD, he hopes the committee will use these new data to help inform clinical decision-making. “Obviously, it’s up to them in terms of how to put it together, but I think now there’s more granularity in terms of what the trade-offs would be, as they decide what level of recommendations to give the procedures,” he said.

If There’s a Difference, It’s Small

Gregg Stone, MD (Icahn School of Medicine at Mount Sinai, New York, NY), who led the EXCEL trial, said the results are in line with what he expected based on a study-level analysis published last year, although the new pooled patient-level data is of much higher quality. Like prior data, the meta-analysis conclusively shows there’s no difference in mortality between PCI with DES and CABG surgery in patients with left main CAD.

“And if there is to be a difference, that difference would be quite small,” he told TCTMD. 

Based on cumulative evidence, including the latest meta-analysis, Stone said that PCI for left main CAD should be a class I indication for cases where operators can safely achieve complete or near complete revascularization. In most instances, that would likely be in patients with low-to-intermediate SYNTAX scores, said Stone.

Similarly, David Kandzari, MD (Piedmont Heart Institute, Atlanta, GA), said the meta-analysis has been positioned as an attempt to resolve the tension and debate over the revascularization of left main CAD, particularly the controversy surrounding mortality and MI risks in the different trials. Led by an independent research group, but including patient-level data, the meta-analysis largely reaffirms the EXCEL data, in which he was an investigator, as well as other meta-analyses.

“My expectation is that this analysis should satisfy societal and guideline position statements,” Kandzari told TCTMD. “Overall, and I’m certainly not part of the guideline writing committee, but I wouldn’t expect this to change the European Society of Cardiology’s endorsement for left main PCI.”

And if there is to be a difference, that difference would be quite small. Gregg Stone

For cardiothoracic surgeon Jack Boyd, MD (Stanford University School of Medicine, CA), the meta-analysis also adds very little to the current debate, although he has a different perspective on the results.

“I don’t think it’s going to change the way that anybody thinks,” he told TCTMD. “From EXCEL and NOBLE, the outcomes are pretty clear. PCI can be done safely, but the outcomes aren’t as good. Patients can choose that, and that’s reasonable, if it’s a high priority to avoid surgery because of some surgical risk or because they’re unwilling to undergo [surgery]. But long-term they’re going to do better with surgery. I don’t think that’s changed.”

Coronary revascularization, said Boyd, is important for the treatment of left main CAD, both for the resolution of symptoms and prognosis. Patients have options, he said, and while cardiologists and surgeons will all look at the same data, they’ll ultimately spin it different ways.

“I think if you did the Martian test or brought somebody in that wasn’t involved—and you have to because I’m a surgeon and my interests are what they are—I’m confident that an impartial person would present the data much differently than it’s being presented,” said Boyd. “I think the conclusions reached would be consistent with the surgeon’s opinion but you’re going to have to go to a non-surgeon and non-interventionalist to get that point.”

Subodh Verma, MD, PhD (University of Toronto, Canada), the Canada Research Chair in Cardiovascular Surgery, doesn’t believe the issue is fully resolved yet, even with the new data. Although the excess risk was nonsignificant, Verma noted there were numerically fewer deaths among those treated with surgery at 5 years.    

“There is a mortality benefit, even in these low-to-moderate risk groups, albeit the magnitude of this is likely small, but over a lifetime [it does] add up in favor of CABG over PCI,” said Verma, adding that surgery should remain the treatment of choice in patients with high SYNTAX scores. Moreover, medical therapy matters. He pointed out that just 41.3% of patients in the surgical arm received a P2Y12 inhibitor compared with 97.9% of those who underwent PCI and a comparison between surgery and PCI in those who received a P2Y12 inhibitor would be warranted.

Given the complexities involved in choosing the right revascularization strategy for patients with left main CAD, Verma emphasized that the heart team approach is truly warranted. “The current meta-analysis exemplifies this and is the most prudent approach, in my opinion, moving forward,” he said. Like Boyd, he congratulated Sabatine and colleagues for conducting an “important and rigorous meta-analysis” in a controversial area of cardiology.

I’m confident that an impartial person would present the data much differently than it’s being presented. Jack Boyd

Speaking with TCTMD, cardiovascular surgeon Vinod Thourani, MD (Piedmont Heart Institute), emphasized the importance of the heart team, pointing out that the meta-analysis includes patients with left main CAD “who could have gone either way” with respect to surgery and PCI.

“It’s like with TAVR,” he said. “To me, it’s analogous to bicuspid aortic stenosis, in some ways. Outcomes can be good with transcatheter technology if the heart team decides there is equipoise in treating that disease’s anatomical process. With bicuspid valves, if the anatomy is good, TAVR does a good job. If the anatomy is bad, TAVR won’t do a good job. Here, if you have good left main coronary artery disease that’s amenable to PCI by a heart team assessment, then at least at 5 years you’re not seeing a significant mortality difference. I think I can walk away saying that to patients.”

The onus, he stressed, is on cardiologists and surgeons to look at these patients together to determine if the anatomical complexity allows for equipoise between the two procedures. Every heart team will need to make decisions based on the quality of their interventionalists and surgeons.

With respect to the revascularization guidelines, including those for patients with three-vessel CAD based on the recent FAME-3 results, Thourani believes they do need to be reassessed, with particular attention paid to the short- and long-term clinical outcomes. In terms of follow-up, Thourani said 10 years would be ideal, but that he’s satisfied with seeing data out to 5 years. “For me, I’m comfortable with about 80% confidence at 5 years that we’re making the right decisions,” said Thourani.

Spontaneous and Procedural MIs

Regarding the higher risk of spontaneous MI, Stone said this finding is explained by differences in the two procedures.   

“It’s most likely due to the fact that bypass surgery bypasses long coronary segments of coronary artery plaque whereas PCI is more of a ‘spot’ treatment of the more severe lesions, although this is somewhat offset by a lower rate of procedural myocardial infarction,” said Stone. “When you put it together, it helps explain why there’s no difference in cardiovascular death because the long-term rates of myocardial infarction are similar.”

To TCTMD, Kandzari said that operators have been aware of the excess risk of MI and repeat revascularization with PCI compared with surgery since the 3- and 5-year data from SYNTAX were published. In contrast, data suggest the risk of stroke could be higher with surgery, at least in the first year. Kandzari said the evidence has largely shown that cardiovascular survival will be similar with PCI and surgery, and “the findings from this meta-analysis don’t change that message.”

He noted that surgeons would need to perform 29 bypass operations to prevent one MI and 14 to prevent one repeat revascularization.

As to the difference in procedural MI risks, Stone said PCI was associated with a lower risk of protocol-defined procedural MI, which should be the primary definition used in the meta-analysis because it was prespecified. Kandzari added that the procedural MI risks will likely be “in the eye of the beholder,” but he too believes the protocol-defined procedural MI events shouldn’t be discounted because these have been shown to be associated with a higher likelihood of death, largely cardiovascular mortality.

We’re bickering a little too much on who won and who didn’t. That’s not the point. Our goal is not to win. Vinod Thourani

Thourani, on the other hand, stressed that CABG surgery should not be “penalized” in clinical trials for the increase in periprocedural MI risk based on cardiac troponin elevations as opposed to those captured by changes on ECG. For that reason, the Universal Definition is preferred over some definitions selected by study investigators, including the modified SCAI version chosen by the EXCEL researchers.

Time to Stop Bickering

Ultimately, though, the meta-analysis should help highlight the risk and benefits of both procedures, as well as the factors that need to be weighed by the heart team when making decisions for individual patients.

“Of course, what’s not in this meta-analysis is differences in quality of life and other procedural complications, such as major bleeding, early rehospitalizations, acute kidney injury, atrial fibrillation requiring chronic oral anticoagulation, all of which are much more frequent after bypass surgery compared with PCI,” said Stone. The short-term benefit of PCI needs to be balanced against fewer spontaneous MIs and repeat revascularization. “It’s really trade-off,” he added. “It’s not that one procedure is better than the other.”

Kandzari, for his part, hopes the meta-analysis will be part of a collaborative presentation to patients from the heart team. The debate between surgeons and patients over the pros and cons of each strategy can’t occur in a vacuum, he added, noting that patient preference must be considered. “What’s meaningful to a doctor might not be meaningful to a patient,” said Kandzari.

Thourani said he is fortunate to have an excellent working relationship with interventional cardiologists, with whom he has a long history of collaboration; his hospital was one of the largest enrollers of patients for EXCEL. Five or six times per week, he is asked to assess different patients where interventionalists have called for a surgical consult. Going forward, he’d like to see equal representation of open-minded surgeons and cardiologists as trialists in the big randomized trials—and that both groups need to be open-minded to the benefits of both procedures—because it’s patients who’ll ultimately benefit.

“That’s what it’s all about,” said Thourani. “Our goal should be the balanced care of each individual patient. I think we’re bickering a little too much on who won and who didn’t. That’s not the point. Our goal is not to win. Our goal is to be able to look at a patient and say this is the right choice for you at this moment in your life.”    

Michael O’Riordan is the Managing Editor for TCTMD. He completed his undergraduate degrees at Queen’s University in Kingston, ON, and…

Read Full Bio
  • Sabatine reports research grant support through Brigham and Women’s Hospital from Abbott, Amgen, Anthos Therapeutics, AstraZeneca, Bayer, Daiichi-Sankyo, Eisai, Intarcia, Ionis, The Medicines Company, MedImmune, Merck, Novartis, Pfizer, and Quark Pharmaceuticals, and has consulted for Althera, Amgen, Anthos Therapeutics, AstraZeneca, Bristol Myers Squibb, CVS Caremark, DalCor, Dr Reddy’s Laboratories, Fibrogen, IFM Therapeutics, Intarcia, MedImmune, Merck, Moderna, Novo Nordisk, and Silence Therapeutics.
  • Stone reports consulting fees/honoraria/speakers bureau fees from Elucid Bio, HeartFlow, Abiomed, Valfix, TherOx, Gore, Ablative Solutions, Miracor Medical SA, Ancora, Vectorius, Robocath, Neovasc, and Cardiomech; equity/stocks/options in SpectraWave, Orchestra Biomed, Applied Therapeutics, Biostar funds, Aria, Cardiac Success, Valfix, Ancora, and Cagent; and grant/research support from Philips, Vascular Dynamics, V-Wave, Biosense Webster, Abbott Vascular, Bioventrix, and CSI.
  • Kandzari reports equity/stock(s)/options (personal) from Biostar Ventures; grant support/research contracts (institutional) from Medtronic, Teleflex, Abbott Vascular, CSI, and Biotronik; and consultant fees/honoraria/speakers bureau fees (personal) from CSI.
  • Thourani reports grant/research support from Abbott Vascular, Edwards Lifesciences, Boston Scientific, and W.L. Gore & Associates.
  • Verma reports no relevant conflicts of interest.