Individual patient-level data meta-analysis of NOBLE, EXCEL, SYNTAX and PRECOMBAT: PCI vs. CABG for patients with left main coronary artery disease | SCAI

Why is this study important for interventional cardiologists?

Most studies comparing PCI and CABG for left main disease have found the two treatments provide similar results when coronary complexity is not severe. In 2019, the European Society for Cardio-Thoracic Surgery (EACTS) withdrew support for European revascularization guidelines out of concern conclusions drawn from randomized trial data did not accurately represent differences in late mortality and MI rates, arguing that results were superior for CABG. The meta-analysis presented at the American Heart Association Scientific Sessions 2021 was intended to clarify the facts around those concerns. As detailed below, a meta-analysis of pooled patient-level data showed that PCI is a reasonable alternative to CABG.

 What question was the study supposed to answer? 

The key question was whether late mortality (after 5 years or more) was different between PCI and CABG for left main disease. Concerns had also been raised about myocardial infarction rates with the two treatments.

What did the study show?

Patient-level data were pooled from 4 randomized studies (NOBLE, EXCEL, SYNTAX Left Main cohort, and PRECOMBAT) to assess outcomes after PCI or CABG. No important difference in all-cause death (PCI: 11.2%, CABG: 10.2%, HR 1.10 [CI 0.91-1.32], p=0.33) or CV death (6.2% vs 5.9%, HR 1.07 [0.83-1.37], p=0.61) at 5 years or all-cause death at 10 years (all-cause death: 22.4% vs 20.4%, HR 1.1 [0.93-1.29], p=0.24) was found. Between 1 and 5 years, all-cause death was numerically greater after PCI but this fell short of statistical importance (p=0.072); this difference was chiefly due to greater non-cardiovascular deaths after PCI. The mortality curves appeared to widen between years 1 and 4, then narrow again between years 4 and 5. A Bayesian probability analysis concluded that the difference observed is likely real but small (e.g., a 49.1% probability that the difference is ≤0.2%/year, and just 5.2% probability that the difference is ≥0.5%/year). So, if a mortality difference does exist, it is sufficiently small that it should not preclude consideration of PCI. The MI data analysis confirmed that the definition used impacts estimated risk: the protocol definition yielded significantly more procedural MIs with CABG and the WHO Universal Definition yielded no important difference in procedural MIs. CABG resulted in fewer spontaneous MIs (6.2% vs 2.6%, HR 2.35 [1.17-3.23], p<0.0001) but not lower rates of cardiovascular death.

These findings suggest that the European guidelines were reasonable and appropriate; new US revascularization guidelines are anticipated in the next few months.

All Editors: Kirk M. Garratt, MD, MSc, MSCAI; and David A. Cox, MD, MSCAI