PCI Equal to Surgery in Left Main CAD: PRECOMBAT at 10 Years

(UPDATED) There is no significant difference in the long-term risk of major adverse cardiovascular or cerebrovascular events (MACCE) among patients with left main CAD treated with PCI or CABG surgery, according to new data from the PRECOMBAT trial.

At 10 years, the primary composite endpoint of all-cause mortality, MI, stroke, or ischemia-driven revascularization occurred in 29.8% of patients treated with PCI and 24.7% of patients who underwent CABG surgery (HR 1.25; 95% CI 0.93-1.69). There was no difference in the risk of death, MI, or stroke in the two treatment arms, but the rate of revascularization was significantly higher in patients treated with PCI (HR 1.98; 95% CI 1.21-3.21).

Presented during the late-breaking clinical trial session at the virtual American College of Cardiology 2020 Scientific Sessions, lead investigator Duk-Woo Park, MD, PhD (Asan Medical Center, Seoul, Korea), said the study was not powered for clinical outcomes at 10 years and the results should be considered hypothesis-generating only. But their data don’t back the findings from EXCEL showing that PCI was associated with an increased risk of all-cause mortality, he observed.  

“The data are still very limited, especially beyond 5 years,” said Park, adding that the “available long-term studies show conflicting results.” Some studies, he said, have shown equivalence between PCI or CABG in short-term follow-up only to see the benefits of CABG surgery emerge over time. Other randomized trials, like EXCEL and NOBLE, showed different results, while the 5-year data from EXCEL ignited a firestorm between surgeons and interventional cardiologists.

The big question is how this trial informs a field that has such controversy and very divergent findings from trial to trial. Frederick Welt

Marc Moon, MD (Washington University School of Medicine in St. Louis, MO), one of the discussants following the presentation, said the absence of differences in the risk of death, MI, or stroke in both the intention-to-treat (ITT) and as-treated analysis should be reassuring to physicians and patients.

“Almost every study of coronary surgical revascularization versus PCI has shown a much higher revascularization rate with PCI,” said Moon. “That’s just one of the things we have to accept for the capability of doing PCI. I think that’s a risk we balance in each patient, and the patient needs to understand it and the treating physician needs to understand it.”

Frederick Welt, MD (University of Utah School of Medicine, Salt Lake City), another panelist during the late-breaking session, said the long-term data are critical because “we’re all concerned about the catch-up phenomenon, where perhaps CABG will tend to have a benefit that manifests at a later time point.” He praised the investigators for their follow-up, noting that 96% of patients in both study arms were tracked in PRECOMBAT. “It’s fantastic for a trial that wasn’t designed to go 10 years,” he said. “The big question is how this trial informs a field that has such controversy and very divergent findings from trial to trial.”

In the light of all of the recent debate, Park acknowledged that sorting through which revascularization strategy is best in the setting of left main CAD is a challenge.

Mortality in PRECOMBAT and Elsewhere

PRECOMBAT was the first randomized study to compare PCI versus CABG in patients with left main disease. Conducted between 2004 and 2009, the study included 600 patients with unprotected left main CAD randomly assigned to PCI with a first-generation sirolimus-eluting stent (Cypher; Cordis) or CABG surgery. Intravascular ultrasound was used in more than 91% of patients treated with PCI, while 93.6% of patients in the CABG arm underwent revascularization of the LAD with an internal-thoracic-artery graft. Baseline characteristics were well matched between the two groups, with 64.6% of patients having distal left main bifurcation disease and 32.0% having diabetes. The mean Syntax score was 24.8, and the trial included 22.3% of patients with a high Syntax score (≥ 33).

Overall, there was no significant differences in the risk of MACCE between the two treatment arms at 10 years, nor was there any significant difference in the risk of death, MI, or stroke (18.2% with PCI vs 17.5% with CABG; HR 1.00; 95% CI 0.70-1.44). Rates of MI—both Q-wave and non-Q-wave MI—did not differ between the two strategies, nor was there any difference in the risk of stroke.

Importantly, they observed no difference in the risk of death from any cause, deaths from cardiovascular causes, or deaths from undetermined causes. In PRECOMBAT, 14.5% of patients treated with PCI had died from any cause at 10 years compared with 13.8% of patients in the surgical arm.

In the as-treated analysis, the MACCE rate was 31.6% in 327 patients treated with PCI and 21.8% in 272 patients treated surgery, a statistically significant difference (HR 1.51; 95% CI 1.11-2.06) that was driven by the increase in target lesion revascularization. There was no difference in the rates of death, stroke, or MI. As treated, 17.2% in the PCI arm required revascularization in follow-up compared with 5.9% in the surgery arm (HR 2.69; 95% CI 1.56-4.62).

Park said that there were some “imbalances” between the two treatment groups in the as-treated analysis so the results should be interpreted cautiously. The treatment effect in the ITT analysis was consistent across multiple subgroups, but PCI was associated with a higher risk of death, MI, stroke, and ischemia-driven target lesion revascularization in patients with three-vessel disease. There was no difference in event rates in the PCI- and CABG-treated patients when stratified by SYNTAX score. 

Making Sense of the Findings

While the issue of increased harm with PCI was raised following the 5-year EXCEL results, Park pointed out the 5-year results from NOBLE didn’t confirm those findings, nor did extended follow-up from the SYNTAX study of patients with left main CAD. Now their analysis has also failed to show any differences between PCI and CABG in terms of all-cause mortality. He said their event rates were lower than in other trials, but that differences in the patient population, as well as the increased use of IVUS, might explain the lower rate.

Speaking with TCTMD, Gregg Stone, MD (Icahn School of Medicine at Mount Sinai, New York, NY), lead investigator of the EXCEL trial, said the all-cause mortality event rates in PRECOMBAT were “virtually superimposable” and in terms of “the three endpoints that really matter—death, MI, and stroke—there was no significant difference between the two procedures.”

Along with the extended follow-up from SYNTAX, which showed the rate of all-cause mortality was nearly equivalent between surgery and PCI with a first-generation paclitaxel-eluting stent (Taxus; Boston Scientific), “both experiences are very reassuring that at 10-year follow-up, patients are not falling off a cliff if they’ve had PCI,” said Stone. “We hear frequently from the surgeons that the benefits of surgery will be experienced with long-term follow-up, and to some extent that’s true, but also with very long-term follow-up grafts start to fail, especially saphenous vein grafts, which are still the most widely used bypass grafts today.”

That said, Stone said that in patients with triple-vessel CAD, PRECOMBAT showed that patients fare better with bypass surgery, a finding widely accepted by cardiologists.

In contrast, Sanjay Kaul, MD (Cedars-Sinai Medical Center, Los Angeles, CA), took issue with the conclusions of the PRECOMBAT investigators, stating they were misleading because the trial was designed as a noninferiority study, and had the investigators applied the same statistical criteria at 10 years as they had a 1 year, it wouldn’t have shown PCI was equivalent to CABG surgery.

For a claim of “no significant difference” to be made, it would ideally be based on a prespecified hypothesis of noninferiority established at 1 year, according to Kaul. At that time, the study met its noninferiority criteria margin but the event rate was lower than expected, which would bias the results towards noninferiority.

At 10 years, there were more than three times those seen at 1 year, and the upper boundary of the difference in risk was 12.4%, which would not have met the “liberal” noninferiority margin of less than 7% at 1 year.

“One could argue that with so many events, a much stricter margin would have been justified,” Kaul wrote in an email. “When one looks at the HR, the confidence interval ranges from a 7% relative risk reduction to a 69% increase with most of the estimates lying above a HR of 1.0. The results, whether expressed in risk difference or HR, favor CABG. When one looks at the results of the as-treated or per-protocol analysis, the more robust platform for noninferiority studies, the results clearly favor CABG.”

For Moon, though, these long-term data start a new conversation in the ongoing debate between PCI versus CABG surgery for the treatment of left main CAD. “What it definitely shows is that if we have a patient who is not a good candidate for coronary revascularization surgically, we can expect an ‘acceptable’ result with PCI in some high-risk patients,” he said, noting that most of the patients in PRECOMBAT had a low-to-intermediate SYNTAX score.

Stone called the SYNTAX score a “relatively blunt instrument,” and while the finding that disease severity was a better predictor of outcomes than the SYNTAX score is an interesting observation, it needs to be teased out further. In EXCEL at 5 years, they did not observe a significant interaction between clinical outcomes and disease severity (number of diseased vessels) or SYNTAX score. He also noted the PRECOMBAT investigators didn’t observe an interaction in patients with and without diabetes, a finding they also reported in EXCEL.

“My own personal feeling is that the diabetes interaction has been misconstrued,” he said. “I believe there is a relationship between diabetes and worse outcomes after PCI compared with CABG, but it’s probably mostly confined to patients who have the most severe coronary anatomy and perhaps other end-organ manifestations of diabetes. But if you have left main disease and noncomplex coronary disease, most of those patients will do just fine with PCI even if they have diabetes.”  

Mario Gaudino, MD (Weill Cornell Medicine, New York, NY), who was not involved in the study, called PRECOMBAT a pioneering clinical trial but pointed out it was conducted more than 15 years ago. EXCEL and NOBLE are more contemporary trials, and the long-term PRECOMBAT data are unlikely to alter the perspective of practicing physicians. “Surgery is associated with better outcomes in the long term, but PCI is a very good alternative for the patient who can’t have surgery, if their risk is too high or if they’re more concerned about short-term outcomes,” he advised.

Note: Stone is a faculty member of the Cardiovascular Research Foundation, the publisher of TCTMD.