Meta-analysis Shows Similar 5-Year Mortality for PCI, CABG in Left Main Disease

The study-level data comes amid major scrutiny of left main trials, culminating in an announcement today about an EXCEL review.

Meta-analysis Shows Similar 5-Year Mortality for PCI, CABG in Left Main Disease

Patients with left main coronary artery disease who undergo PCI are at no higher risk of dying in long-term follow-up when compared with individuals treated with CABG surgery, according to the results of a new study-level meta-analysis.

After a mean follow-up of more than 5.5 years, patients with left main CAD treated with PCI and CABG surgery had comparable rates of mortality, as well as similar rates of cardiac death, stroke, or myocardial infarction, although the rate of unplanned coronary revascularization was higher among the PCI-treated patients.  

These findings, say researchers, may help quiet some of the controversy that sprung up following the publication of 5-year outcomes from the EXCEL trial. In that study, the risk of all-cause mortality was significantly higher in patients treated with left-main PCI versus CABG surgery. This led to a series of heated exchanges between surgeons and interventional cardiologists about the merits of the two revascularization procedures in this setting—and calls for an independent review of EXCEL.

The new meta-analysis was published March 2, 2020, in the European Heart Journal. Two days later, the BMJ reported that the New England Journal of Medicine has in fact launched a review of EXCEL.

“Cardiologists and surgeons alike want to be data-driven, and we both want to be able to offer the best treatment for patients,” lead investigator on the EHJ paper, Yousif Ahmad, MD (NewYork-Presbyterian/Columbia University Irving Medical Center, New York, NY), told TCTMD. “The mortality results from our meta-analysis are reassuring for us as cardiologists, for surgeons, and for the patients. The reason being is that the point estimates for all-cause mortality are bang in the middle. It’s not like we’re seeing a trend in one direction or the other.”

I’m hoping that a paper like ours might help settle down the narrative somewhat and make it more dispassionate and data-driven. Yousif Ahmad

Given that there was no risk of cardiac mortality with PCI in EXCEL, as well as no excess risk of cardiac deaths in the meta-analysis, their results can help aid clinical decision-making when counseling patients with left main CAD, said Yousif.

“If the heart team decides that the patient is an appropriate candidate for either mode of revascularization—it’s important that this initial decision-making and consensus is taken collaboratively—then we can present the patient with the best data we have,” said Ahmad. “That is: up to about 5 years of follow-up, on average, the survival with both treatments appears to be the same. One treatment is less invasive than the other, but it may result in further treatments down the line. Depending on the patient’s values and preferences, they have the freedom to choose between the two knowing there’s not a real penalty for one or the other.”

Gregg Stone, MD (Icahn School of Medicine at Mount Sinai, New York, NY), one of the meta-analysis’ authors and lead investigator of EXCEL, made a similar argument.

“I believe that with follow-up out to 5 to 10 years, you can expect comparable mortality for most patients eligible for both procedures,” he told TCTMD. “Then there are also all of the other differences the meta-analysis doesn’t cover, such as a more rapid return to work, better early quality of life, less periprocedural complications, and so on.”     

Reappraising the LM Landscape 

The meta-analysis includes 4,612 patients with left main coronary artery disease randomized to PCI or CABG surgery in EXCEL, NOBLE, SYNTAX, PRECOMBAT, and a small German study.

To TCTMD, Ahmad said the new analysis wasn’t a response to the current controversy that started when David Taggart MD, PhD (University of Oxford, England), the chairman of the EXCEL surgical committee during the design and recruitment phase of the trial, criticized the trial leadership for downplaying the increased risk of all-cause mortality and not publishing complete MI data. Instead, it came about following the recent publication of extended follow-up from the three largest studies of left main revascularization, including the 5-year follow-up from EXCEL and NOBLE as well as the 10-year results from SYNTAX.

“We thought it appropriate to reappraise all of the data with the longer-term results, especially when you’re comparing surgery with stenting where longer-term data becomes very apposite,” said Ahmad. “That was the real motivation rather than in response to controversies or perceived controversies with any one trial.”

The results showed there was no difference in the risk of all-cause mortality between PCI with drug-eluting stents and CABG surgery (RR 1.03; 95% CI 0.82-1.30). Nor was there a difference in the risk of cardiac mortality (RR 1.03; 95% CI 0.79-1.34). In a sensitivity analysis that excluded EXCEL, there was again no significant difference in the risk of all-cause mortality between the two procedures, but the “moderate” heterogeneity between studies evident in the overall analysis was no longer seen.

There was also no difference in the risk of stroke between PCI and surgery (RR 0.74; 95% CI 0.36-1.50), although here “substantial” heterogeneity was documented. Stone noted that NOBLE showed an “odd” finding of increased stroke between 1 and 5 years with PCI, a finding which hasn’t been observed in any other study. “When NOBLE was removed [from the data set], then there was a clear reduction in stroke with PCI compared with CABG that has been seen in almost all other comparative trials,” said Stone.  

There was no difference in the risk of MI (RR 1.22; 95% CI 0.96-1.56), but in studies with procedural and nonprocedural MI data reported, procedural MI was more common in patients treated with surgery but nonprocedural MI more frequent in PCI-treated patients. The risk of unplanned coronary revascularization was significantly higher with PCI compared with CABG surgery (RR 1.73; 95% CI 1.49-2.02).

Nick Curzen, MD, PhD (University Hospital Southampton NHS, England), who wasn’t involved in the present study, said that despite the limitations inherent to meta-analyses, it provides comfort that there is likely no difference in mortality in carefully selected patients with left main coronary artery disease treated with PCI or CABG surgery.

“It does provide additional reassurance that the all-cause mortality difference we saw at 5 years in EXCEL probably shouldn’t lead to a concern about PCI having a higher risk of mortality in the left main than bypass surgery,” said Curzen. “For me, that’s what this paper offers, but I’m a great fan of papers that provide one simple message and that’s the one simple message this paper offers.”

Based on the new data, it appears that the increased risk of all-cause mortality observed in EXCEL was a “chance finding,” said Curzen. “That’s a very valuable thing for us to see.” While the higher rate of unplanned revascularization with PCI is not news, it’s also an important message that physicians should convey to patients eligible for either revascularization procedure.

Curzen pointed out that patients randomized in the various trials met strict inclusion criteria, which can make real-world decision-making challenging. “Strictly speaking, the patients we see in real life, quite a proportion of them—and it varies depending on the study we’re talking about—would not have been in the randomized trial, and we’re left extrapolating data from a different cohort of patients to the real world,” he said. “We all need to be aware of that.”

Patient-Level Meta-analysis Coming Soon

To TCTMD, Ahmad said MI is a slightly less robust endpoint than mortality for a study-level meta-analysis, the reason being that they had to use each trial’s definition for periprocedural and procedural events. “Overall, the results are more or less as we’d expect,” he said. “Not all trials reported procedural and nonprocedural MIs very exhaustively, so our available data for those endpoints are less. But I think they show more or less what we’d expected: that procedural MI seems to be greater after surgery, while nonprocedural MI, particularly late, seems to be greater with PCI.”

The EXCEL investigators, in fact, have come under a lot of scrutiny for not reporting all of their MI data, specifically the rate of procedural MI adjudicated using the Third Universal Definition, which was a prespecified secondary endpoint. The group has said that cardiac troponin levels were not routinely collected during the trial, which limited their ability to analyze MI rates using this definition, but that they plan to publish further studies on MI rates in the two study arms.

On the whole, Ahmad said he hopes the meta-analysis will provide yet more evidence supporting collaboration between the different specialties, noting that all randomized patients in these trials were reviewed by surgeons and interventionalists. He also hopes the data will help calm the waters.

“Within that context of collaborative work and the heart team, I was surprised to see the degree of response to EXCEL and the way it snowballed slightly in terms of the tone taken during the discussions,” he said. “I’m hoping that a paper like ours might help settle down the narrative somewhat and make it more dispassionate and data-driven.”  

One of the limitations of the present analysis, as acknowledged by the researchers, is that it provides only aggregate outcome data. Stone said they are currently working on a patient-level meta-analysis of the four major randomized trials of left-main DES PCI versus CABG surgery. The advantage of the patient-level meta-analysis is that it provides data on temporal relationships, such as time-to-event curves, allows researchers to look at subgroups, and identify multivariable predictors of events.

“We’re working on that, and hopefully that will be out this summer,” said Stone. 

In December 2019, following a BBC Newsnight report of the missing MI data in the EXCEL trial, as well as allegations the researchers ignored safety concerns from the data safety and monitoring board, the European Association of Cardio-Thoracic Surgery withdrew their support for the current treatment recommendations for left main coronary artery disease. Curzen, who is the president of the British Cardiovascular Intervention Society, said they their group continues to adhere to the European guidelines, noting they issued a statement of support following the BBC report.

Note: Stone as well as several other co-authors of the meta-analysis are faculty members of the Cardiovascular Research Foundation, the publisher of TCTMD.

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
  • Ahmad reports no relevant conflicts of interest.
  • Stone reports receiving speaker or other honoraria from Cook, Terumo, QOOL Therapeutics and Orchestra Biomed; serving as a consultant to Valfix, TherOx, Vascular Dynamics, Robocath, HeartFlow, Gore, Ablative Solutions, Miracor, Neovasc, V-Wave, Abiomed, Ancora, MAIA Pharmaceuticals, Vectorious, Reva, Matrizyme; and receiving equity/options from Ancora, Qool Therapeutics, Cagent, Applied Therapeutics, Biostar family of funds, SpectraWave, Orchestra Biomed, Aria, Cardiac Success, MedFocus family of funds, Valfix.