Research in context
Evidence before this study
We searched MEDLINE, Embase, and the Cochrane Library up to July 19, 2017, to identify randomised clinical trials comparing coronary artery bypass grafting (CABG) with percutaneous coronary intervention (PCI) using stents. We used the search terms “coronary artery bypass grafting”, “percutaneous coronary intervention”, “stent”, and “random*”. Studies were included if the patients had multivessel or left main coronary artery disease and did not present with acute myocardial infarction, PCI was done with bare-metal or drug-eluting stents and not balloon angioplasty, and more than 1 years' follow-up for all-cause mortality was available. We identified 12 high-quality trials. One trial found a survival benefit of CABG over PCI with bare-metal stents for multivessel disease at 6 years' follow-up. Another trial found better survival at 5 years' follow-up with CABG than with PCI using first-generation drug-eluting stents in patients with multivessel disease and diabetes. However, these results have not been reproduced in other individual trials with 3–10 years' follow-up, except in underpowered and hypothesis-generating subgroup analyses. Two pooled analyses of CABG versus PCI with balloon angioplasty or bare-metal stents for multivessel disease found conflicting results, and what the survival differences are between CABG and PCI remains largely unclear.
Added value of this study
This study is the largest analysis of patients randomly assigned to PCI using stents or to CABG. To our knowledge, this study shows for the first time that all-cause mortality is significantly lower with CABG than with PCI in an overall randomised population of patients with multivessel or left main coronary artery disease. Additionally, the use of individual patient data allowed identification of important subgroups that have a survival benefit from CABG. These subgroups include patients with multivessel disease and diabetes and those with higher coronary lesion complexity (established with the Synergy between PCI with Taxus and Cardiac Surgery [SYNTAX] score). Patients with left main disease had similar survival with PCI and CABG, regardless of diabetes and SYNTAX score.
Implications of all the available evidence
Some patients have specific indications for PCI or CABG, such as coronary complexity too high for PCI or operative risk too high for CABG. In patients with estimated clinical equipoise, as determined by heart teams, consideration of disease type (multivessel or left main), coronary complexity, and diabetes status is crucial because these are important treatment effect modifiers of favourable mortality after CABG versus PCI and should affect decisions on coronary revascularisation in daily practice. However, longer follow-up of randomised trials is needed to better define mortality differences in overall patients and specific subgroups.