News

PCI vs. CABG debate continues in left main after meta-analysis shows tradeoffs, no clear winner

Concrete PCI superiority, noninferiority to CABG in LMCAD remains undeclared; meta-analysis suggests no mortality difference but tradeoffs on MI, stroke, repeat revasc

A clear-cut answer regarding the optimal revascularization strategy for left main coronary artery disease (LMCAD) has yet to emerge, even after a recent comprehensive meta-analysis and four landmark randomized controlled trials published over the past decade.

The fierce, unresolved debate over whether percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) is the superior treatment strategy for CAD patients with left main disease raged on following the recent presentation of a meta-analysis at the virtual American Heart Association's (AHA) 2021 Scientific Sessions.

Lead investigator Marc S. Sabatine, MD (Brigham Women's Hospital, Boston, USA) presented the findings that hinged principally on the four landmark SYNTAX[1] [2], PRECOMBAT[3] [4], NOBLE[5] [6], and EXCEL[7] [8] studies that compared the two revascularization strategies. The study was conducted as a collaboration between six independent investigators - including Sabatine - and the principal investigators of the SYNTAX, PRECOMBAT, NOBLE, and EXCEL trials.

The team found no statistically significant difference for 5-year all-cause mortality between PCI and CABG among 4,394 LMCAD patients (PCI 11.2% vs. CABG 10.2%, HR 1.10, 95% CI, 0.91-1.32, p=0.33). However, Bayesian analysis showed an 87.5% probability that 5-year death was greater with PCI, albeit the difference being less than 0.2% per year and pertaining more to non-cardiovascular deaths.

5-year Kaplan-Meier estimate rates for mortality between PCI and CABG
Type of death PCI CABG Difference (HR, 95% CI)
CV 6.2% 5.9% 0.4 (-1.1 to 1.8)
Non-CV 5.2% 4.5% 0.7 (-0.6 to 2.0)

Sabatine noted that the cumulative incidence curves for cardiovascular mortality did not progressively diverge over time. The all-cause mortality rates between PCI and CABG compared to the all-cause mortality of the revasc strategies in SYNTAX and PRECOMBAT - two trials with 10-year follow-up - were also similar (CABG 22.1% vs. PCI 21.6%; HR 0.96, 95% CI, 0.76-1.21, P=0.72).

There were tradeoffs for each approach. Investigators found that spontaneous myocardial infarction (MI) risk was higher with PCI than CABG (6.2% vs. 2.6%; HR 2.35; 95% CI 1.71-3.23, p<0.0001). PCI also had higher risk for revascularization than CABG (18.3% vs. 10.7%; HR 1.78, 95% CI, 1.51-2.10, p<0.0001).

From the beginning, it depended on individual patient characteristics and preferences on what they are trying to avoid or achieve.

Robert Harrington, MD

There was no difference regarding stroke risk (2.7% vs. 3.1%, HR 0.84, 0.59-1.21, p=0.36), but PCI had a lower risk of stroke within the first year following randomization (HR 0.37; 0.19-0.69).

Sabatine and colleagues concluded in the Lancet paper[9] simultaneously published on Nov 15 that: "Among LMCAD patients with largely low or intermediate coronary artery complexity, there was no statistically significant difference in 5-year all-cause mortality between PCI and CABG, although a Bayesian approach suggested a difference probably exists (<0.2% per year) favoring CABG."

"There were tradeoffs regarding the risk of MI, stroke, and revascularization," the team wrote. "A heart team approach to communicate expected outcome differences can help patients reach a treatment decision."

In a related Lancet editorial[10], Emilie P. Belley-Côté pointed out: "This result raises the question of what the noninferiority threshold or allowed maximum excess risk of death would be for researchers to declare PCI as non-inferior to CABG."

"For the outcome of mortality, PCI's superiority or noninferiority to CABG remains unproven," Belley-Côté said. "Given that PCI is less invasive and requires a shorter recovery time than CABG, many interventional cardiologists have viewed PCI as a compelling alternative for LMCAD."

"But for PCI to become the new golden standard, it must prove superior to CABG. For PCI to be a credible alternative, it must show noninferiority to CABG," she added. "Although the [individual patient data meta-analysis] research team deserves appreciation for their efforts, the data shows that the totality of patients randomly assigned across all trials to date is insufficient to clearly establish superiority or noninferiority of PCI over the gold standard of CABG."

A trial larger than the aggregate of SYNTAX, PRECOMBAT, NOBLE, and EXCEL should be a priority in the field.

Emilie P. Belley-Côté, MD

Belley-Côté critiqued that the meta-analysis focused only on superiority, which was not established for PCI, and did not attempt to directly inform PCI's noninferiority by setting a noninferiority boundary - although physicians could "draw their own conclusions" given the 85.7% probability that PCI had a higher 5-year death risk and a 49.1% probability the excess absolute risk was 1% or greater.

"A trial larger than the aggregate of SYNTAX, PRECOMBAT, NOBLE, and EXCEL should be a priority in the field," she said. "Until definitive trial data become available, patients deserve to be presented with the advantages, disadvantages, and uncertainties of both approaches [by] cardiac surgeons and interventional cardiologists."

Junghoon Lee, MD (Asan Medical Center, Seoul, South Korea) told SummitMD: "The subjective nature of anatomic complexity and process of narrowing down CABG candidates - also affected by the volume and experience of the medical center - demands careful analysis by the heart team when selecting a revascularization strategy."

"This meta-analysis, like the PRECOMBAT trial, found no statistically significant difference between PCI and CABG" Lee said. "Although certain study limitations exist - such as the average SYNTAX score of 25 and 22% of patients showing high anatomic complexity (33) - these findings pave the way forward to the creation of individualized strategies for many patients."

Little to no mortality diff between PCI-CABG eases some concerns, sparks others

Persistent uncertainty on the optimal revascularization strategy was fueled by several landmark trials, including the EXCEL trial, that set different composite endpoints and showed inconsistent findings, as previously reported by SummitMD[11].

Although limitations exist, these findings pave the way forward to the creation of individualized strategies for many LMCAD patients.

Junghoon Lee, MD

Based on the investigations for LMCAD, current European and US guidelines recommend CABG (Class I) for all LMCAD patients regardless of anatomical complexity.

The 2018 European Society of Cardiology and European Association for Cardio-Thoracic Surgery (ESC/EACTS) joint guideline[12] recommends PCI for LMCAD with a Class I indication for LMCAD with a low SYNTAX score and a Class IIa for LMCAD-intermediate SYNTAX score. European guidelines advise against PCI for LMCAD with high SYNTAX scores (33).The recent ACC/AHA/SCAI joint guideline recommended CABG for most stable ischemic heart disease patients with severe left main stenosis and high anatomic complexity CAD (COR: 1; LOE: B-R). PCI was suggested for patients without high anatomic complexity and unsuitable for CABG (COR 2a; LOE: B-NR).

Despite guidelines, the simmering discord in the cardiology community erupted into a public standoff at the end of 2019 between EXCEL principal investigator Gregg W. Stone, MD (Icahn School of Medicine at Mount Sinai, New York, USA) and cardiac surgeon David Taggart, MD, PhD (University of Oxford, Oxford, England) - who voluntarily withdrew his authorship from the publication citing problems with trial design and results.

The latest meta-analysis - expected to be a "tie-breaker" of sorts - fell short, although it provided insight with the added Bayesian analysis for the ongoing debate regarding PCI and CABG, commentators said.

A tie may not be broken, but we now have in front of us the truth - and that allows for conversations between heart teams and patients.

Roxana Mehran, MD

AHA 2021 discussant Robert Harrington, MD (Stanford University, California, USA) said: "While I would like to say this meta-analysis breaks the tie, it's more realistic to say we have more opportunity to explore the question with great quantitative data."

"Before EXCEL, the 2017 ACC guidelines already pointed out that choosing between PCI and CABG is not a simple yes or no question, requiring both anatomic considerations and careful consideration of patient preference," Harrington said. "From the beginning, there was no difference between bypass or PCI. Instead, it depends on individual patient characteristics, values and preferences on what the patient is trying to avoid or achieve."

Sabatine added: "Stroke weighs heavily in people's minds, and this can lead to difficult conversations that may sound like how many strokes versus nonfatal MIs would you like to avoid.' It's a complex decision."

Amidst humorous comments on involving a statistician to explain trial results during patient and heart team conversations, panelist Roxana Mehran, MD (Mount Sinai School of Medicine, New York, USA) said soberingly that the "unbiased" meta-analysis does better than break a tie - it reveals the "truth" of the matter.

"This is an independent analysis that involved the original investigators but removed the bias," Mehrane said. "Although a tie may not be broken, we have the truth in front of us - and that allows for conversations between heart teams and patients based on data."


  1. https://www.nejm.org/doi/full/10.1056/nejmoa0804626
  2. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(19)31997-X/fulltext
  3. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(19)31997-X/fulltext
  4. https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.120.046039
  5. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(16)32052-9/fulltext
  6. https://www.thelancet.com/article/S0140-6736(19)32972-1/fulltext
  7. https://www.nejm.org/doi/full/10.1056/nejmoa1610227
  8. https://www.nejm.org/doi/full/10.1056/nejmoa1909406
  9. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(21)02334-5/fulltext
  10. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(21)02491-0/fulltext
  11. https://summitmd.com/html/technical/news_view.php?cmode=daily&no_case=48
  12. https://academic.oup.com/eurheartj/article/40/2/87/5079120

Leave a comment

Sign in to leave a comment.