SAN DIEGO, CA—There is no significant difference in the risk of all-cause mortality at 10 years in real-world patients with left main coronary artery disease treated with PCI or coronary artery bypass graft (CABG) surgery, according to results of the IRIS-MAIN registry.
Risks of spontaneous MI and target-vessel revascularization were higher among the PCI-treated patients, but the risk of stroke was lower when compared with surgical group, reported investigators today at TCT 2026.
Duk-Woo Park, MD (Asan Medical Center, Seoul, South Korea), who led the registry analysis, said that “for patients suitable for either strategy, [a] multidisciplinary heart team discussion might help patients and clinicians weigh potential trade-offs in the choice between PCI and CABG.”
In the registry, which was published simultaneously in the Lancet, the estimated 10-year mortality rate among the 1,543 patients treated with CABG surgery was 22.9% and 24.0% among the 3,836 patients who received a drug-eluting stent in the overlap-weighted analysis. The difference between the two groups was not statistically significant (weighted hazard ratio 1.00; 95% CI 0.87-1.14).
“The reason for that is, by definition, the folks that are selected for surgery tend to be a lower-risk population,” he said. “They’re not as frail, they don’t have as many comorbidities, and so on,” said Kirtane. “Almost uniformly, when registries are done like this, they’ll show a worse outcome, especially on the mortality side, for PCI.”
In contrast, surgical patients in IRIS-MAIN were considerably higher risk the PCI-treated patients.
The findings line up with a recent meta-analysis of randomized trials comparing PCI to CABG in patients with left-main coronary artery disease, said Patrick Serruys, MD, PhD (University of Galway, Ireland).
Ajay Kirtane, MD (NewYork-Presbyterian/Columbia University Irving Medical Center, NY), who moderated a morning press conference, said the results are very encouraging because registry data tends to disadvantage PCI.
In one, which was presented at EuroPCR 2026 and is still unpublished, all-cause mortality at 10 years was 23.5% in those who underwent surgery and 23.1% in those treated with PCI. The difference was not statistically significant. A participant-level meta-analysis, published early this year in JACC, also showed nearly identical rates of all-cause mortality.
“It’s interesting that the randomized trials preceded the registry but came up with the same [mortality],” said Serruys.
The Backstory, In Brief
Several years ago, considerable debate erupted around the risks of mortality between the two revascularization strategies, a controversy launched following the 5-year data from the EXCEL trial. In that trial, PCI was noninferior to CABG surgery for the study’s primary endpoint of all-cause mortality, MI, or stroke, but all-cause mortality was significantly higher among the PCI-treated patients (13.0% with PCI vs 9.9% with surgery; OR 1.38; 95% CI 12.03-1.85).
After EXCEL, the European Association for Cardio-Thoracic Surgery (EACTS) withdrew its support of the revascularization guidelines, which in turn led the European Society of Cardiology (ESC) to ask the TIMI Study Group to review the relevant data.
The TIMI meta-analysis, which included SYNTAX, PRECOMBAT, NOBLE, and EXCEL, showed there was no difference in the risk of mortality with CABG surgery and PCI at 5 years, but the risk of spontaneous MI and coronary revascularization were higher with PCI. The 10-year follow-up, which focused only on mortality because data were not available on other endpoints, also showed no difference in risks of death with the two revascularization strategies.
Evidence is still lacking from a real-world population, especially an Asian population.
Duk-Woo Park, MD, PhD
Following the review, an ESC/EACTS task force updated the revascularization guidelines to state that PCI should be a class IIb indication in patients with left main disease of low-to-intermediate anatomic complexity. In the US, CABG surgery is recommended (class 1, level of evidence B) to improve survival for stable ischemic heart disease patients with significant left main stenosis. PCI is a class 2a recommendation for selected patients with left main disease and a low-to-intermediate SYNTAX score.
“The contemporary guideline is largely based on randomized trials in a highly selective cohort of patients,” said Park. “Evidence is still lacking from a real-world population, especially an Asian population, with complex anatomy and treated with widespread imaging-guided PCI.”
IRIS-MAIN Registry
The IRIS-MAIN registry, with sites in India, Malaysia, Japan, Indonesia, and South Korea, included patients with unprotected left main coronary disease eligible for revascularization with PCI or CABG surgery.
At baseline, those treated with surgery were more likely to have diabetes, be current smokers, and have a history of MI, stroke/transient ischemic attack, peripheral artery disease, chronic obstructive pulmonary disease, and congestive heart failure. They also were more likely to have reduced left ventricular ejection fraction. Regarding anatomy, CABG-treated patients more commonly had disease in the left main plus three-vessel disease. They were also more likely to have distal bifurcation lesions.
In all, 81.1% of patients underwent PCI guided by intravascular ultrasound and 83.0% received a second-generation or newer stent. In the surgery arm, 94.1% received a left internal thoracic artery graft, 46.3% received an additional radial artery graft, and 28.5% were treated only with arterial grafts. Off-pump CABG was performed in 59.4% of the surgical cases.
Regarding 10-year secondary endpoints, spontaneous MI occurred in 4.6% of patients treated with PCI and 1.9% of those treated with surgery (subdistribution HR 2.06; 95% CI 1.15-3.66). Stroke rates were 3.0% and 6.2% in the PCI- and CABG-treated patients, respectively (sHR 0.53; 95% CI 0.33-0.86). Rates of target-vessel revascularization were 11.8% in the PCI-treated patients and 4.6% in those treated with surgery (sHR 2.57; 95% CI 1.74-3.79).
Multiple sensitivity analyses using different statistical models confirmed the overall findings. The main results were also consistent across prespecified subgroups, including in patients with diabetes mellitus.
Park, during the late-breaking clinical trials session, acknowledged the potential for selection bias with the registry data. He also pointed out that the SYNTAX score, which is used to assess anatomic complexity, was not prospectively collected by sites. Their findings are observational, he said, and shouldn’t be misconstrued to suggest any “causal treatment effects.”
Usually a Disadvantage to PCI
During the discussion following the presentation, panelists noted that the quality of surgery at participating sites was excellent, with more than one-quarter of patients receiving complete arterial revascularization.
Yousif Ahmad, MD, PhD (University of California, San Francisco), noted that registries from North America and Europe show that higher risk patients are more likely to be treated with PCI. Also, these same registries suggest that CABG surgery is still the first choice for coronary revascularization in left main disease. For example, only 15% of 11,137 patients with left main coronary artery disease in the SWEDEHEART registry were treated with PCI while just 5.8% of 22,586 patients in a Canadian registry underwent PCI.
Park pointed out that 33 sites from South Korea participated in the IRIS-MAIN registry, and these operators have considerable expertise in left-main PCI. In the last 20 years, PCI of left main disease in Korea is increasingly done, he said.
“It’s been the norm for a while in Korea, and it’s partly cultural, but when there’s equipoise, there’s a large preference for PCI,” agreed TCT panelist Kyung Woo Park, MD, PhD (Seoul National University Hospital, South Korea), during the discussion.
Source:
Wee S-B, Kang D-Y, Ahn J-M, et al. Long-term outcomes after percutaneous coronary intervention with drug-eluting stents versus coronary artery bypass grafting for left main coronary artery disease: a prospective analysis of the multicenter IRIS-MAIN registry. Lancet. Published online November 3, 2026.
Disclosures:
Duk-Woo Park reports grant/research support from Abbott, Boston Scientific, Daiichi Sankyo/Eli Lilly, Edwards Lifesciences, and Medtronic.
Kirtane reports grant support/research contracts from Abbott, Medtronic, Boston Scientific, Amgen, CathWorks, Siemens, Philips, Recor Medical, Spectranetics, Chiesi, Concept Medical, Neurotronik, Biotronik, Magenta, Canon, SoniVie, Shockwave Medical, Abiomed, Teleflex, and Supira. He reports consulting for SoniVie.
Kyung Woo Park reports consulting fees/honoraria from Daiichi-Sankyo/Eil Lilly, Shockwave Medical, Amgen, Novartis, Samjin Pharmaceutical, and HK InnoN Pharmaceutical.
Serruys reports grant/research support from Merillife and Novartis.
Ahmad reports consulting fees/honoraria from Boston Scientific, Abbott, Shockwave Medical, Cardiovascular Systems, Edwards Lifesciences, and Verge Medical.