In defense of the EXCEL Trial: PCI vs. CABG for left main disease

February 26, 2020

Historically, CABG was the only option for left main coronary disease. With improvements in stent technology and procedural skill, interventional cardiologists were able to safety treat left main disease with percutaneous techniques. The EXCEL trial compared PCI to CABG in patients with low or intermediate anatomic complexity left main disease. The trial has become a flashpoint of controversy, now many months after its final five-year report appeared in the New England Journal of Medicine.

With completion of five-year follow-up of this RCT, the authors concluded that there was no significant difference between PCI and CABG in the composite of death, stroke, or myocardial infarction. Like almost all composite endpoints in other trials, each component was given equal weight. In other words, a myocardial infarction was counted equally to death or stroke. An asymptomatic myocardial infarction might have triggered the primary endpoint: equal in its statistical consequences to a devastating stroke or a death. While this fact in and of itself can be problematic, similar designs have been utilized and accepted in countless other trials.

What have been the criticisms of this study? The news outlets were somewhat late to the game, now six months after the article appeared. But after the European Association of Cardiothoracic Surgery recently withdrew its support for guidelines on the treatment of left main disease, recent posts by the BBC and other outlets were quick to identify some potential shortcomings of the trial. These limitations included controversies and disagreements related to an alleged change in the definition of myocardial infarction midway through the trial, conflicts of interest related to investigators and authors, and contentions that the data safety monitoring board raised concerns about the increased risk of death in the PCI group during the course of the trial. The debate is ongoing.

This much we know. First, a composite endpoint was necessary to limit the sample size to a reasonable number. Even then, it took 126 sites in 17 countries and 3.5 years to complete enrollment. Second, the mortality difference in favor of CABG was not surprising; this has been a common finding in other studies of PCI vs CABG. While not entirely comparable, in a review of five RCTs of multivessel coronary disease, late mortality was increased by approximately 50% after PCI compared with CABG (Spadaccio C, Benedetto U. Ann Cardiothorac Surg 2018;7:506-15). Third, the authors are among the most experienced clinical trialists in the field. No clinical trial is perfect, and while there can be many confounders even in RCTs, the design and execution of this study appears as good as any.

The real question that remains, however, is whether patients would accept a slightly higher mortality to avoid an open surgical procedure. This is a question that can only be addressed in the confines of the pre-procedure consultation between the doctor and the patient. The answer will vary and there is no uniformly correct answer.

Controversies over the EXCEL trial results will persist, and no doubt be debated for quite some time. However, the take-home message from the EXCEL trial doesn’t change. PCI is remarkably effective for left main disease, and many patients will choose to tolerate a slightly higher rate of long-term complications to avoid a median sternotomy.

Stone GW, Kappetein AP, Sabik JF, Pocock SJ, Morice M, et al., for the EXCEL Trial Investigators*. Five-Year Outcomes after PCI or CABG for Left Main Coronary Disease. N Engl J Med 2019; 381:1820-1830